# Airbnb

Creative engineers and data scientists building a world where you can belong anywhere. http://airbnb.io - Medium

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## De-identifying text at scale: How Airbnb protects personal data in AI training

DevFeed: [De-identifying text at scale: How Airbnb protects personal data in AI training](<https://devfeed.tech/articles/de-identifying-text-at-scale-how-airbnb-protects-personal-data-in-ai-training-138584.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/de-identifying-text-at-scale-how-airbnb-protects-personal-data-in-ai-training-32f3ff8ff7d0?source=rss----53c7c27702d5---4>)

Author: Alexandr Climov

Published: 2026-11-17T18:01:03Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [pii](<https://devfeed.tech/topics/pii.md>), [privacy-filter](<https://devfeed.tech/topics/privacy-filter.md>), [managing large-scale text data in AI](<https://devfeed.tech/topics/managing-large-scale-text-data-in-ai.md>), [Data Quality](<https://devfeed.tech/topics/data-quality.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-training](<https://devfeed.tech/tags/ai-training.md>), [airbnb](<https://devfeed.tech/tags/airbnb.md>), [data](<https://devfeed.tech/tags/data.md>), [de-identification](<https://devfeed.tech/tags/de-identification.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [extensibility](<https://devfeed.tech/tags/extensibility.md>), [hosting](<https://devfeed.tech/tags/hosting.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [messages](<https://devfeed.tech/tags/messages.md>), [modularity](<https://devfeed.tech/tags/modularity.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [personal-data](<https://devfeed.tech/tags/personal-data.md>), [pii](<https://devfeed.tech/tags/pii.md>), [presidio](<https://devfeed.tech/tags/presidio.md>), [privacy](<https://devfeed.tech/tags/privacy.md>), [reviews](<https://devfeed.tech/tags/reviews.md>), [support](<https://devfeed.tech/tags/support.md>), [technology](<https://devfeed.tech/tags/technology.md>), [text](<https://devfeed.tech/tags/text.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

Airbnb de-identifies text used for model training with an on-premises Presidio library integrated into its shared ML workflow. The Analyzer detects and labels PII; the Anonymizer resolves overlapping spans and replaces them. Airbnb added phrase-based context matching for company-specific identifiers and uses GLiNER for PERSON and LOCATION because spaCy's broad location tags did not reliably capture full street addresses in its benchmark. Detection thresholds and entity scope vary by consumer. Most text PII receives plausible Faker-generated replacements; per-run secret seeding keeps repeated values consistent within a run while changing fakes across runs, without storing a real-to-fake map. Structured data uses salted hashes for identifiers, truncated coordinates, and format-preserving encryption for internal IDs. Benchmarks report recall, precision, and throughput by entity type and language; releases must avoid regressions. The library's protection depends on pipelines using a current version and compliant configuration, and new languages require labeled data for evaluation.

### Source excerpt

Here's how Airbnb detects and replaces personal data in messages, reviews, and support chats before it reaches a model. By: Emelin Flores, Alex Klimov, Lee Formyduval, Bahador Nooraei Introduction Airbnb is built on trust. Every night that a guest books a stay, or a host welcomes someone into their home, that trust is backed by information they've shared with us, which includes personally identifiable information (PII) such as names, home location, current location, and payment details. We now use transaction information to train machine learning models. Before we do so, we remove PII to minimize the number of copies of this information and prevent leakage into and through our models. As we've written before, Airbnb maintains a shared taxonomy as to what constitutes PII, from commonly used fields such as names and card numbers to less frequently used fields such as precise location coordinates. We use de-identification to remove PII and replace it with simulated data that's as realistic as possible to maximize its usefulness for ML model training. Decisions relating to user privacy start as policy. These policies are set and maintained by Airbnb's Privacy Architecture team, with engagement from Legal, Policy, and Engineering. Engineering then implements policy decisions as code. This post focuses on text de-identification. We'll describe the OSS library we chose to start with, the changes we made to create a functional system for Airbnb, and how we confirm that the system is working. 1. Choosing Presidio as the foundation Many organizations are concerned with protecting PII, so there's a lot of existing best practice and tested code out there that we can use. We chose Presidio, an open-source library originally developed by Microsoft and recently transitioned to community governance. We needed to meet demanding requirements: On-premises hosting, to keep data on our infrastructure General-purpose functionality and high degree of modularity, for maximum extensibility

## Skipper, Airbnb's embedded workflow engine, is now open source

DevFeed: [Skipper, Airbnb's embedded workflow engine, is now open source](<https://devfeed.tech/articles/skipper-airbnb-s-embedded-workflow-engine-is-now-open-source-127298.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/skipper-airbnbs-embedded-workflow-engine-is-now-open-source-78dd354f472b?source=rss----53c7c27702d5---4>)

Author: Andriy Sergiyenko

Published: 2026-11-10T18:01:02Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [workflow automation](<https://devfeed.tech/topics/workflow-automation.md>), [Open Source](<https://devfeed.tech/topics/open-source.md>), [Flowable](<https://devfeed.tech/topics/flowable.md>)

Tags: [airbnb](<https://devfeed.tech/tags/airbnb.md>), [apache](<https://devfeed.tech/tags/apache.md>), [apache-2-0](<https://devfeed.tech/tags/apache-2-0.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [business-processes](<https://devfeed.tech/tags/business-processes.md>), [classes](<https://devfeed.tech/tags/classes.md>), [crashes](<https://devfeed.tech/tags/crashes.md>), [database](<https://devfeed.tech/tags/database.md>), [debugging](<https://devfeed.tech/tags/debugging.md>), [dependencies](<https://devfeed.tech/tags/dependencies.md>), [dependency](<https://devfeed.tech/tags/dependency.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [documentation](<https://devfeed.tech/tags/documentation.md>), [embedded](<https://devfeed.tech/tags/embedded.md>), [engine](<https://devfeed.tech/tags/engine.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [skipper](<https://devfeed.tech/tags/skipper.md>), [software-architecture](<https://devfeed.tech/tags/software-architecture.md>), [technology](<https://devfeed.tech/tags/technology.md>), [workflow](<https://devfeed.tech/tags/workflow.md>), [workflow-engine](<https://devfeed.tech/tags/workflow-engine.md>)

### AI overview

Skipper is an open-source Java/Kotlin workflow engine that runs in a service's own replicas and uses its existing database, avoiding a separate orchestration tier between workflow steps. A shared database stores state and scheduled work; replicas poll for tasks and use leases so another replica can recover work after a failure. This suits teams that want durable workflows inside their existing deployment, but they take responsibility for execution, database capacity, and recovery. Workflows replay from the beginning, reusing persisted action results, so orchestration code must be deterministic; actions should make side effects idempotent because effects and checkpoints are not atomic and may run at least once. Checkpoints are normally batched, reducing database writes but allowing completed actions to run again after a crash before persistence. The Apache 2.0 release includes MySQL and SQLite adapters, test utilities, an admin UI, and a Kotlin state machine DSL.

### Source excerpt

Durable workflows, running inside your service with no separate orchestration cluster. By: Ricardo Gamba, Andriy Sergiyenko Skipper, an embedded workflow engine used in production by teams all across Airbnb, is now available to the open source community. Skipper lets you express multi-step business processes as Java or Kotlin classes, persist their progress, and recover from crashes and transient failures. It runs inside your service and reuses its existing dependencies. Skipper is available under the Apache 2.0 license. The code is on GitHub, alongside documentation and runnable examples. Our earlier engineering deep dive explains how we built it. Here, we'll focus on the architecture of Skipper and where it fits in the ecosystem of options for development around workflows. Why we developed Skipper A process that spans multiple services, waits for human input, or involves a mix of synchronous and asynchronous steps, needs persistent storage for recording its progress. When a server restarts, it must know what has already happened, what (if anything) should be retried, and what is still awaiting processing. Implementing this separately for each process would mean repeatedly solving the same problems around partial failure, timers, and recovery, with associated testing and debugging; using a proven library is likely to improve quality and reduce toil. Existing workflow engines meet these challenges. A common approach relies on a central orchestration service, whether self-hosted or fully managed, to handle durable execution. This setup provides a unified platform for multiple teams, streamlining central operations and monitoring. For Airbnb, we wanted a new approach, one in which each service owned its workflows within its existing deployment and database. For critical services, that meant avoiding an additional shared orchestration dependency. It also let workflow execution stay in-process, without a separate coordinator between steps. That requirement shaped Skippe

## CASTLE: training natural language search before the first real query

DevFeed: [CASTLE: training natural language search before the first real query](<https://devfeed.tech/articles/castle-training-natural-language-search-before-the-first-real-query-86238.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/castle-training-natural-language-search-before-the-first-real-query-3818ef8b4828?source=rss----53c7c27702d5---4>)

Author: Wendy Ran Wei

Published: 2026-10-13T17:01:02Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [search-engines](<https://devfeed.tech/topics/search-engines.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [airbnb](<https://devfeed.tech/tags/airbnb.md>), [cold-start-problem](<https://devfeed.tech/tags/cold-start-problem.md>), [data](<https://devfeed.tech/tags/data.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [few-shot](<https://devfeed.tech/tags/few-shot.md>), [generate](<https://devfeed.tech/tags/generate.md>), [generation](<https://devfeed.tech/tags/generation.md>), [language-model](<https://devfeed.tech/tags/language-model.md>), [large-language-model](<https://devfeed.tech/tags/large-language-model.md>), [llm](<https://devfeed.tech/tags/llm.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [model](<https://devfeed.tech/tags/model.md>), [natural-language](<https://devfeed.tech/tags/natural-language.md>), [relevance](<https://devfeed.tech/tags/relevance.md>), [search](<https://devfeed.tech/tags/search.md>), [synthetic](<https://devfeed.tech/tags/synthetic.md>), [technology](<https://devfeed.tech/tags/technology.md>), [traffic](<https://devfeed.tech/tags/traffic.md>), [train](<https://devfeed.tech/tags/train.md>), [training](<https://devfeed.tech/tags/training.md>), [training-data](<https://devfeed.tech/tags/training-data.md>)

### AI overview

CASTLE tackles cold-start natural language search by using de-identified booking sessions to form positive and negative listing pairs, then prompting an LLM to generate queries that fit the booked listing but not the alternatives. This makes relevance labels part of the construction, avoiding a separate LLM judging step for the core data. Template prompts, about 500 research seed queries, attribute-focused prompts, and a variety mechanism are used to improve query variety and match guest phrasing. For listings without booking history, a Virtual Judge assigns labels; CASTLE-R is intended for settings with some real search traffic. In the reported evaluation, CASTLE achieved 1.01 KL divergence against real query distributions, versus 9.33 for the best baseline, and experts agreed with its labels 91-93% of the time across 200 sampled triplets. These results are reported for the described evaluations.

### Source excerpt

How we addressed the cold-start problem for Airbnb's natural language search by jointly generating synthetic training data -- queries and relevance labels -- from de-identified booking and search signals. By: Wendy Ran Wei, Hao Li, Weiwei Guo, Xiaowei Liu, Xueyin Chen, Chun How Tan, Malay Haldar, Soumyadip Banerjee, Kedar Bellare, Huiji Gao, Sanjeev Katariya, and Stephanie Moyerman Introduction Natural language search lets guests express intent in their own words, such as "romantic cabin near skiing" or "pet-friendly beach house that sleeps eight," rather than navigating filters and categories. We needed to introduce this kind of search for Airbnb, and to have it work well from the start. However, this required training the model before we had actual user queries to train the model on, and a reasonable set of relevance labels (inferred from the queries) to map queries to matching listings. Without prior traffic, both must be synthesized from scratch. This is the cold-start problem for natural-language search. Other organizations have faced similar challenges and have created methods to address them. However, prior approaches address the need for a body of queries and the need for relevance labels separately. Passage-based methods such as InPars [1] prompt a large language model (LLM) to generate queries conditioned on a single document, then rely on additional LLM judgments to assign relevance labels, introducing a second source of error. Promptagator [2] improves query diversity using few-shot examples, but is still limited by the gap between its generation distribution and real user behavior. In this post, we describe Contrastive And Seed-guided Training for natural Language sEarch (CASTLE), our framework for jointly generating queries and producing relevance labels from Airbnb's own de-identified booking data. Our paper describing CASTLE was accepted as a full research paper at the 35th ACM International Conference on Information and Knowledge Management, more wide

## Beyond synthetic testing: Capturing and replaying real database workloads at Airbnb

DevFeed: [Beyond synthetic testing: Capturing and replaying real database workloads at Airbnb](<https://devfeed.tech/articles/beyond-synthetic-testing-capturing-and-replaying-real-database-workloads-at-airbnb-65699.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/beyond-synthetic-testing-capturing-and-replaying-real-database-workloads-at-airbnb-cea7ee9b1ab2?source=rss----53c7c27702d5---4>)

Author: Zuofei Wang

Published: 2026-10-06T17:01:04Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [Transactions](<https://devfeed.tech/topics/transactions.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [database-consistency](<https://devfeed.tech/topics/database-consistency.md>), [clickhouse](<https://devfeed.tech/topics/clickhouse.md>)

Tags: [airbnb](<https://devfeed.tech/tags/airbnb.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [behavior](<https://devfeed.tech/tags/behavior.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [capture](<https://devfeed.tech/tags/capture.md>), [challenges](<https://devfeed.tech/tags/challenges.md>), [client](<https://devfeed.tech/tags/client.md>), [clusters](<https://devfeed.tech/tags/clusters.md>), [complete](<https://devfeed.tech/tags/complete.md>), [critical](<https://devfeed.tech/tags/critical.md>), [custom](<https://devfeed.tech/tags/custom.md>), [data](<https://devfeed.tech/tags/data.md>), [database](<https://devfeed.tech/tags/database.md>), [databases](<https://devfeed.tech/tags/databases.md>), [debug](<https://devfeed.tech/tags/debug.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [error](<https://devfeed.tech/tags/error.md>), [fast](<https://devfeed.tech/tags/fast.md>), [fleet](<https://devfeed.tech/tags/fleet.md>), [framework](<https://devfeed.tech/tags/framework.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [sql](<https://devfeed.tech/tags/sql.md>), [synthetic](<https://devfeed.tech/tags/synthetic.md>), [technology](<https://devfeed.tech/tags/technology.md>), [testing](<https://devfeed.tech/tags/testing.md>)

### AI overview

Airbnb built a client-agnostic database traffic capture and replay system on ProxySQL to test real MySQL-compatible workloads offline. ProxySQL logs are moved to object storage; a processor separates traffic by cluster, restores transaction order, buckets queries into five-minute files, and pins captured auto-increment IDs in INSERTs. That rewrite keeps dependent reads consistent across database versions, but means replay does not test the target's native ID generation. In "Replay only" mode, traffic runs against one target at configurable speeds or QPS, with errors, latency, and resource use tracked for load testing and capacity planning. "Replay and compare" runs the same traffic against two targets and records result differences for upgrade and migration checks. Workers preserve query ordering, relative timing, and concurrency. Airbnb reports using replays to find MySQL 5.7-to-8.0 latency regressions and queries with nondeterministic ordering; scaling one cluster's write traffic by 80% raised average commit latency from about 6 ms to 34 ms. Captured data is treated as production data, and replays run only in production-equivalent environments.

### Source excerpt

How we capture real production database traffic at Airbnb and replay it offline to load-test, plan capacity, and de-risk upgrades. By: Zuofei Wang, Erluo Li Introduction At Airbnb, MySQL-compatible databases are a critical backbone of our online database infrastructure: a fleet of hundreds of clusters supporting thousands of use cases at millions of queries per second (QPS). Operating databases at scale brings hard problems, including sizing clusters for future growth, keeping behavior consistent across version upgrades and migrations, and reproducing production incidents well enough to debug them. This post describes the database traffic capture and replay system we built to help us tackle them. Challenges and motivation A complex database operation such as Airbnb's requires a number of supporting systems, and one of them is a way to capture database queries and replay them. This is needed for error recovery, governance, quality assurance testing and product improvement efforts. Previously, Airbnb's system for database query capture and replay was fragmented. Each language binding used its own logging framework to emit query traces to Kafka, which were then processed per use case and replayed by a custom replayer. Legacy Query Capture and Replay System This system was borrowed from our observability pipeline. It worked as a fast initial solution, but over time, we found that it fell short of what we needed. Capturing queries client-side and per language carried a heavy maintenance burden whose ownership was ambiguous, and it scaled poorly across our service-oriented architecture (SOA). Most importantly, it lacked enough information to convey a complete picture of each transaction. As a result, it couldn't accurately replay transactions, which made it impossible to verify that the same queries return the same results on the new database. To close these gaps, we set out to build a system for query capture and replay, with three goals: Load testing with real productio

## Personalization without user identity

DevFeed: [Personalization without user identity](<https://devfeed.tech/articles/personalization-without-user-identity-61983.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/personalization-without-user-identity-8e7891a1d486?source=rss----53c7c27702d5---4>)

Author: Wei Jiang

Published: 2026-09-29T17:01:01Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [personalization](<https://devfeed.tech/topics/personalization.md>), [recommendation systems](<https://devfeed.tech/topics/recommendation-systems.md>)

Tags: [acquisition](<https://devfeed.tech/tags/acquisition.md>), [advertising](<https://devfeed.tech/tags/advertising.md>), [aggregate](<https://devfeed.tech/tags/aggregate.md>), [ai](<https://devfeed.tech/tags/ai.md>), [airbnb](<https://devfeed.tech/tags/airbnb.md>), [blog](<https://devfeed.tech/tags/blog.md>), [blog-post](<https://devfeed.tech/tags/blog-post.md>), [booking](<https://devfeed.tech/tags/booking.md>), [browsing](<https://devfeed.tech/tags/browsing.md>), [but](<https://devfeed.tech/tags/but.md>), [can](<https://devfeed.tech/tags/can.md>), [cases](<https://devfeed.tech/tags/cases.md>), [challenge](<https://devfeed.tech/tags/challenge.md>), [class](<https://devfeed.tech/tags/class.md>), [cold-start-problem](<https://devfeed.tech/tags/cold-start-problem.md>), [collective](<https://devfeed.tech/tags/collective.md>), [core](<https://devfeed.tech/tags/core.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [essential](<https://devfeed.tech/tags/essential.md>), [identity](<https://devfeed.tech/tags/identity.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [personalization](<https://devfeed.tech/tags/personalization.md>), [technology](<https://devfeed.tech/tags/technology.md>), [user](<https://devfeed.tech/tags/user.md>)

### AI overview

Airbnb's Proximity Features address cold-start personalization by using collective activity from nearby users instead of persistent individual identifiers. A proximity key groups roughly 1,000 users: dense areas are subdivided with IP hash buckets, while sparse areas are grouped through progressively coarser geographic tiles. Daily features capture recent engagement, longer-term booking patterns, and bucket metadata. At inference, an IP is mapped to a key and features are fetched from a distributed store; lookup is a soft dependency, so a timeout does not block the request. The design excludes users without consent, uses coarse group-level location, and supports consent and deletion controls. Production experiments report directional gains for never-booked and dormant users in Homepage AutoSuggest, with more region-relevant destinations; marketing landing pages also gained personalized recommendations. Email campaign validation is underway. The text does not provide numerical experiment results.

### Source excerpt

How Airbnb uses proximity signals to personalize without relying on individual user history. By: Wei Jiang, Bin Xu, Bharathi Thangamani, Weiwei Guo, Sundar Srinivasavaradhan, Tracy Yu, Huiji Gao, Swapnil Ghike, Michael Kinoti Great personalization starts with knowing your user. But what happens when the user is a stranger? A significant share of Airbnb users arrive without a login, without a recent search history, or without any prior booking -- especially those landing from paid advertising or organic search. For these users, the ML models that power search ranking, destination recommendations, and marketing landing pages face the cold-start problem: there is little to no user signal available. In this blog post, we describe Proximity Features -- a new class of ML features that addresses both the cold-start problem and evolving privacy constraints simultaneously. The core hypothesis: aggregate activity patterns within a proximity bucket can provide useful signals for cold-start personalization without relying on individual user identity. By grouping nearby users into buckets and aggregating their collective signals, we can construct rich, location-aware features available immediately for any user whose IP can be geo-located -- no login, no prior history, and no persistent user identifier required. The challenge: serving users without individual signals Personalization at Airbnb has historically relied heavily on a user's identifier to link together past searches, bookings, and browsing patterns. This works well for returning, logged-in users. But on marketing and acquisition surfaces, many users are logged out or visiting for the first time, and standard user identifiers are unavailable. In other cases, users are logged in, and standard user identifiers are available, but the features derived from them are stale or too sparse to carry meaningful signals. Privacy regulations compound the challenge. Under GDPR, non-essential tracking for marketing personalization requir

## The guest journey, updated in real time: extending Airbnb's sequence recommender with Chronon

DevFeed: [The guest journey, updated in real time: extending Airbnb's sequence recommender with Chronon](<https://devfeed.tech/articles/the-guest-journey-updated-in-real-time-extending-airbnb-s-sequence-recommender-with-chronon-42165.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/the-guest-journey-updated-in-real-time-extending-airbnbs-sequence-recommender-with-chronon-8f1582578553?source=rss----53c7c27702d5---4>)

Author: Pengyu Hou

Published: 2026-09-17T17:01:02Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [streaming data analytics](<https://devfeed.tech/topics/streaming-data-analytics.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [Event-Streaming](<https://devfeed.tech/topics/event-streaming.md>), [real-time](<https://devfeed.tech/topics/real-time.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [airbnb](<https://devfeed.tech/tags/airbnb.md>), [batch](<https://devfeed.tech/tags/batch.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [feature](<https://devfeed.tech/tags/feature.md>), [journey](<https://devfeed.tech/tags/journey.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [ranking](<https://devfeed.tech/tags/ranking.md>), [real-time](<https://devfeed.tech/tags/real-time.md>), [search](<https://devfeed.tech/tags/search.md>), [sequence](<https://devfeed.tech/tags/sequence.md>), [technology](<https://devfeed.tech/tags/technology.md>), [transformer](<https://devfeed.tech/tags/transformer.md>)

### AI overview

Airbnb replaced daily batch updates to its guest sequence embeddings, which could leave them nearly two days stale, with a Chronon streaming pipeline that typically updates them in under a minute and often within 10-30 seconds. Push Mode notifies downstream consumers when activity features change; Near-real-time Model Transform runs the already-deployed Transformer encoder in that streaming flow. The pipeline merges new activity with existing guest sequence data and writes embeddings to the same low-latency store used by search ranking. Offline evaluation found a 1.67% NDCG improvement over the batch baseline, and online A/B tests found about a one-third of a percent increase in uncancelled bookings. Airbnb says both capabilities are general-purpose Chronon features contributed to its public repository.

### Source excerpt

How two new Chronon capabilities, Push Mode and NRT Model Transform, allows us to provide more relevant search results instantly as a guest explores, rather than waiting for the next batch run. By: Pengyu Hou, Yuli Han, Daochen Zha, Haozhen Ding, Xin Liu, Sophie Wang, Pallavi Adusumilli, Sherry Li, Henry Saputra, Chun How Tan, Huiji Gao, Yan Zhang, Stephanie Moyerman, Yi Li, and Sanjeev Katariya A guest's interaction with Airbnb doesn't pause to wait for a nightly batch job. Someone might browse a dozen listings on a Tuesday afternoon, run a new search that evening, and expect the next search to reflect the recent activity; it's also to Airbnb's benefit for that to be the case. In our previous post, Personalizing Airbnb search by learning from the guest journey, we described how we built a Transformer-based sequence encoder that creates better, more personalized search rankings for a guest using the booking, review, and browsing data that is most relevant to them -- their own. That system ran as a daily batch job: each night it processed the previous day's activity and refreshed embeddings for guests who had something new to show for it. That design worked well, but it left a gap. Activity from earlier the same day wouldn't show up in the embedding until the following day's run, on top of the pipeline's own processing lag -- in practice, up to nearly two days of staleness. For a guest actively planning a trip, that meant the ranking model was often working from a slightly outdated picture of what they wanted, and the recent activities are often highly relevant to current search needs. This is a limitation that our original JourneyFormer research had already flagged as needing new serving infrastructure to solve. In this post, we describe how we closed that gap by adding two new capabilities to Chronon, Airbnb's feature platform: Near-real-time Model Transform and Push Mode. Chronon is an open source project, and these capabilities have been contributed back to our pub

## Beyond the model: Engineering AI infra with scientific judgement

DevFeed: [Beyond the model: Engineering AI infra with scientific judgement](<https://devfeed.tech/articles/beyond-the-model-engineering-ai-infra-with-scientific-judgement-26973.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/beyond-the-model-engineering-ai-infra-with-scientific-judgement-371316d43261?source=rss----53c7c27702d5---4>)

Author: AirbnbEng

Published: 2026-09-15T17:06:18Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [data-engineering](<https://devfeed.tech/topics/data-engineering.md>), [AI for customer service automation](<https://devfeed.tech/topics/ai-for-customer-service-automation.md>), [audit](<https://devfeed.tech/topics/audit.md>), [dataset](<https://devfeed.tech/topics/dataset.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agent-harness](<https://devfeed.tech/tags/agent-harness.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-infra](<https://devfeed.tech/tags/ai-infra.md>), [airbnb](<https://devfeed.tech/tags/airbnb.md>), [analyze](<https://devfeed.tech/tags/analyze.md>), [audit](<https://devfeed.tech/tags/audit.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [conversations](<https://devfeed.tech/tags/conversations.md>), [customer](<https://devfeed.tech/tags/customer.md>), [customer-service](<https://devfeed.tech/tags/customer-service.md>), [customer-support](<https://devfeed.tech/tags/customer-support.md>), [data](<https://devfeed.tech/tags/data.md>), [data-science](<https://devfeed.tech/tags/data-science.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [encoding](<https://devfeed.tech/tags/encoding.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [model](<https://devfeed.tech/tags/model.md>), [science](<https://devfeed.tech/tags/science.md>)

### AI overview

Airbnb's Insight Miner packages a repeatable method for exploring unstructured text around an AI agent. It combines extract, embed, and cluster workflows with prompt tuning, hard-example mining, contrastive labeling, and a path from unsupervised exploration to classification. Researchers start from a question in chat; the harness handles scalable analysis while experts inspect ambiguous or strategic data, test groupings, and build qualitative and quantitative understanding. Airbnb reports that investigations supporting expansion of its customer service assistant to new languages and countries went from months to days, with rigor and speed both increasing. A UI also lets nontechnical subject matter experts run scaled analyses. The excerpt says teams use the tool for survey responses, model evaluation, and fraud-pattern analysis, but gives no comparative benchmark or detailed evaluation results.

### Source excerpt

How Airbnb's agent harness transforms unstructured data exploration by encoding scientific methodology into scalable, reproducible, and audit-ready infrastructure. By: Wren Dougherty Ask a coding agent to analyze 100,000 customer support conversations and within minutes you'll have a polished taxonomy, precise prevalence numbers, and an executive-ready summary. What you can't see is the investigation that produced them: the methods it chose, the evidence it weighed, how much to trust it, or whether a second request would agree. All that reaches you is the polish. The model is undeniably intelligent, but intelligence without methodology is not science. LLMs certainly make for confident scientists, but we need them to be responsible ones. Smarter models help, but intelligence has never been the whole of science, in people or in machines. The method is as much the product as the answer. That is the idea behind the agent harness we built for data science: the methodology itself, built as infrastructure around the model. It governs how an AI agent operates, from framing a question to selecting evidence to recording decisions, so results can be reproduced, audited, and challenged, and the method shared, inspected, and built on. The challenge of unstructured data exploration In 2025, Airbnb was preparing to launch an AI customer service assistant. Before it could ship, we needed to understand exactly what kinds of situations it would face in the real world. That included rare events that could be risky for AI to interact with, and involved examining their taxonomy and prevalence to create the datasets that would help us build a more responsible product. The investigative work to do this was rigorous, but the process was deeply artisanal. Months of high-touch iteration went into each investigation, from finding the right data, reviewing samples with experts, and generating representative datasets, and the method was manually curated across notebooks, tables, docs, and indiv

## Project Lighthouse -- Part 3: Introducing project-lighthouse-anonymize

DevFeed: [Project Lighthouse -- Part 3: Introducing project-lighthouse-anonymize](<https://devfeed.tech/articles/project-lighthouse-part-3-introducing-project-lighthouse-anonymize-1220.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/project-lighthouse-part-3-introducing-project-lighthouse-anonymize-74f8b26653fb?source=rss----53c7c27702d5---4>)

Author: Adam Bloomston

Published: 2026-08-25T17:01:02Z

Content type: release

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [Open Source](<https://devfeed.tech/topics/open-source.md>), [Library](<https://devfeed.tech/topics/library.md>), [Algorithms](<https://devfeed.tech/topics/algorithms.md>), [scikit-learn](<https://devfeed.tech/topics/scikit-learn.md>), [project release](<https://devfeed.tech/topics/project-release.md>)

Tags: [2020](<https://devfeed.tech/tags/2020.md>), [2025](<https://devfeed.tech/tags/2025.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [anonymity](<https://devfeed.tech/tags/anonymity.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [code](<https://devfeed.tech/tags/code.md>), [core](<https://devfeed.tech/tags/core.md>), [data](<https://devfeed.tech/tags/data.md>), [data-quality](<https://devfeed.tech/tags/data-quality.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [implementation](<https://devfeed.tech/tags/implementation.md>), [introduction](<https://devfeed.tech/tags/introduction.md>), [library](<https://devfeed.tech/tags/library.md>), [lighthouse](<https://devfeed.tech/tags/lighthouse.md>), [measurement](<https://devfeed.tech/tags/measurement.md>), [measuring](<https://devfeed.tech/tags/measuring.md>), [methodology](<https://devfeed.tech/tags/methodology.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [opt-out](<https://devfeed.tech/tags/opt-out.md>), [privacy](<https://devfeed.tech/tags/privacy.md>), [published](<https://devfeed.tech/tags/published.md>), [technology](<https://devfeed.tech/tags/technology.md>)

### AI overview

Airbnb open-sourced project-lighthouse-anonymize, a Python library for anonymizing data used to measure potential disparities in user experiences. It applies k-anonymity, with an optional p-sensitive k-anonymity perturbation step, and builds on Core Mondrian with parallel processing, NaN-pattern pre-partitioning, and dynamic suppression budgets. A companion data-quality framework uses Pearson correlation, RILM, and NMIv1, with thresholds empirically validated on synthetic datasets to help assess whether anonymized data can support statistically valid analysis. The article says users can opt out of the underlying data use through account Privacy settings; it provides no comparative evaluation against other anonymization tools.

### Source excerpt

Project Lighthouse -- Part 3: Introducing project-lighthouse-anonymizeThe data in Project Lighthouse is powered by privacy-preserving anonymization code. We've put this code into open source, and published two new technical papers detailing the scalable algorithms and data quality frameworks behind it. By: Adam Bloomston Introduction In 2020, we launched Project Lighthouse, which we developed in partnership with leading civil rights and privacy organizations. As our 2020 announcement details, Project Lighthouse enables us to measure potential disparities in user experiences. This work uses perceived race data that is never linked to individual accounts; we only use this data for measuring potential disparities, and users who want to opt-out can do so by turning off the data use settings in their account's Privacy page. Our results, shared in 2024, demonstrate how we use these analyses to measure our progress in mitigating those disparities. Earlier this year, we open-sourced project-lighthouse-anonymize, the Python library that powers Project Lighthouse's anonymization process. To provide the full technical foundation for this work, we also published two new papers on arXiv alongside the code release. Together with our original 2020 paper, these three papers form a complete story: the foundational methodology, the scalable implementation, and the quality validation framework. The foundational methodology (2020) Our original 2020 paper established the privacy-by-design approach for Project Lighthouse and provides the rationale for choosing k-anonymity as the technical privacy model to prevent sensitive attribute disclosure at scale. For an introduction to this paper, see our first blog post on p-sensitive k-anonymity and our second blog post on measurement with anonymized data. Core Mondrian: Scalable partition-based anonymization (2025) The first of our new papers, Core Mondrian: Basic Mondrian beyond k-anonymity, presents the k-anonymity algorithm at the heart of th

## How we knew COVID was over (and what our models had to unlearn)

DevFeed: [How we knew COVID was over (and what our models had to unlearn)](<https://devfeed.tech/articles/how-we-knew-covid-was-over-and-what-our-models-had-to-unlearn-1218.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/how-we-knew-covid-was-over-and-what-our-models-had-to-unlearn-c606b9bdb0ab?source=rss----53c7c27702d5---4>)

Author: Harrison Katz

Published: 2026-08-19T17:01:03Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [Finance](<https://devfeed.tech/topics/finance.md>), [model architecture](<https://devfeed.tech/topics/model-architecture.md>), [dataset](<https://devfeed.tech/topics/dataset.md>)

Tags: [airbnb](<https://devfeed.tech/tags/airbnb.md>), [bias](<https://devfeed.tech/tags/bias.md>), [cancellations](<https://devfeed.tech/tags/cancellations.md>), [covid](<https://devfeed.tech/tags/covid.md>), [data](<https://devfeed.tech/tags/data.md>), [data-modeling](<https://devfeed.tech/tags/data-modeling.md>), [data-science](<https://devfeed.tech/tags/data-science.md>), [demand](<https://devfeed.tech/tags/demand.md>), [easy](<https://devfeed.tech/tags/easy.md>), [features](<https://devfeed.tech/tags/features.md>), [financial](<https://devfeed.tech/tags/financial.md>), [forecast](<https://devfeed.tech/tags/forecast.md>), [forecasting](<https://devfeed.tech/tags/forecasting.md>), [leave](<https://devfeed.tech/tags/leave.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [markets](<https://devfeed.tech/tags/markets.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>), [rebuild](<https://devfeed.tech/tags/rebuild.md>), [risk](<https://devfeed.tech/tags/risk.md>), [routine](<https://devfeed.tech/tags/routine.md>), [science](<https://devfeed.tech/tags/science.md>), [technology](<https://devfeed.tech/tags/technology.md>)

### AI overview

Airbnb's forecasting team treats a persistent miss as a choice among refitting, respecifying, or holding a model. Refit when the data-generating process remains the same and parameters have drifted; respecify when the model's structure cannot represent changed conditions; hold when a miss falls within expected variation, especially when an unexplained surprise could distort an off-cycle update. Routine refits can absorb temporary spikes or preserve outdated assumptions, while rebuilding in response to a short-lived shock can create a fragile model. During COVID recovery, markets that had once moved together diverged, making the old geographic pooling structure unstable: refitting alone produced about three times the pre-shock baseline error. Changing the model to borrow across geographic neighbors with shared recovery patterns cut error by a little over half versus refitting and brought it close to the pre-shock baseline. The article's central lesson is to reassess and sometimes remove assumptions after a shock, since updating data alone does not correct stale model structure.

### Source excerpt

When we retrain, when we rebuild, and when we leave a model alone. By: Harrison Katz A forecast that carries weight The Forecasting Data Science team at Airbnb produces many of the forecasts the rest of the company plans around: demand, bookings, cancellations, and a range of finer cuts by market and segment, refreshed continuously across thousands of markets. The targets differ, and the models differ, but they have one thing in common: Other teams build on top of them. This means a forecast that is casually wrong is not a clean miss, as it might be in an academic setting. That's because a small bias does not stay small once a lot of decisions are riding on it. So when one of those forecasts starts to drift, what to do about it is not really a methods question. It is a risk question, and an easy one to get wrong, which we have from time to time. One of these forecasts had been missing, compared to what actually happened after the forecast was released, in the same direction for a couple of quarters. This bias persisted after several routine refreshes. The usual solution would be to fully retrain the model: pull in the recent data, refit the model again, and ship. But we wanted to understand the source of the bias, rather than simply hoping an update would eliminate it. If you're interested in other posts on this topic, you can learn more about how COVID impacted Airbnb's financial models or how we dealt with disruption to our models during the pandemic. This post is about the discipline that came out of both: how we now decide whether a struggling forecast needs new data, a new model, or no changes at all. One word, three decisions The easy mistake is treating the choice to "retrain" a model as a single action. It is three separate actions -- refitting, respecifying, or holding -- and none of them is particularly similar to the others. Refitting is the cheaper option. Same model, same structure, same features, updated with newer data. This is what most people mean by

## Flexible Authentication: Reimagining authentication for millions of users at Airbnb

DevFeed: [Flexible Authentication: Reimagining authentication for millions of users at Airbnb](<https://devfeed.tech/articles/flexible-authentication-reimagining-authentication-for-millions-of-users-at-airbnb-1216.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/flexible-authentication-reimagining-authentication-for-millions-of-users-at-airbnb-3a8a4c917137?source=rss----53c7c27702d5---4>)

Author: Jose Santos

Published: 2026-08-12T17:01:04Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [Single sign-on (SSO)](<https://devfeed.tech/topics/sso.md>), [Access Control](<https://devfeed.tech/topics/access-control.md>), [sessions](<https://devfeed.tech/topics/sessions.md>), [User interface design](<https://devfeed.tech/topics/ui-design.md>), [openid](<https://devfeed.tech/topics/openid.md>), [Internationalization (i18n)](<https://devfeed.tech/topics/i18n.md>)

Tags: [airbnb](<https://devfeed.tech/tags/airbnb.md>), [authentication](<https://devfeed.tech/tags/authentication.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [software-architecture](<https://devfeed.tech/tags/software-architecture.md>), [software-engineering](<https://devfeed.tech/tags/software-engineering.md>), [technology](<https://devfeed.tech/tags/technology.md>)

### AI overview

Airbnb's Flexible Authentication flow identifies the account first, then uses a server-side policy engine to choose a likely-to-work verification challenge based on account and session context. Users can switch through a server-ranked "Try another way" list without restarting. Server-defined screens leave clients to render steps, enabling regional changes and experiments without client releases when no client changes are needed. Airbnb reports a 60% code reduction, 100KB smaller web bundle, 20+ experiments in three months, a 2.6% rise in successful authentication, 27% fewer duplicate accounts, and roughly 11% lower SMS OTP costs. The policy engine remains under development.

### Source excerpt

Rebuilding login and signup surfaced product insights, not just technical challenges. Here's how we designed Flexible Authentication at the intersection of product intuition and technical architecture. By: Jose Santos, Mike Barry For Airbnb, logins at irregular intervals are normal. A guest books a trip in January and may not open the app again until summer. A host checks back only when a reservation comes in, and may be busy with other activities when it does. For a two-sided marketplace where a failed login means a lost booking, and lost revenue for both the guest and the host, long gaps between sessions are a structural challenge, not an edge case. Our authentication system had grown organically over a decade, adding new login methods like Social Login, Email OTP, and Phone over time. We recognized that our users have diverse needs: many, such as hosts, use Airbnb daily, while others visit less frequently, when they are planning trips or traveling. To provide a seamless experience, we needed to ensure we were doing a great job supporting all of these usage patterns, helping users pick up exactly where they left off, regardless of how much time had passed or which device they were using. In this blog post, we'll walk through how we rebuilt Airbnb's authentication flows using a new paradigm we call Flexible Authentication, and how each architectural decision was guided by insights into the user's experience. Identify first then Challenge The 'Identify first then Challenge' model was the product insight that reframed the project. Our old system treated authentication as a single question: can this person prove who they are? The real question is more nuanced: given the multiple aspects of the user session, what's the easiest way for them to verify it? A traveler in Brazil who registered with a phone number is better served by a WhatsApp one-time password (OTP) than by SMS, since WhatsApp penetration far exceeds SMS in this market. A returning host in South Korea is b

## Eval-driven development: Lessons from evaluating GenAI at scale

DevFeed: [Eval-driven development: Lessons from evaluating GenAI at scale](<https://devfeed.tech/articles/eval-driven-development-lessons-from-evaluating-genai-at-scale-1215.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/eval-driven-development-lessons-from-evaluating-genai-at-scale-e817e5ae5788?source=rss----53c7c27702d5---4>)

Author: Rohit Girme

Published: 2026-07-28T17:01:03Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [Generative AI](<https://devfeed.tech/topics/generative-ai.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Software Engineering](<https://devfeed.tech/topics/software-engineering.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Developer Tools](<https://devfeed.tech/topics/developer-tools.md>)

Tags: [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [ai](<https://devfeed.tech/tags/ai.md>), [article](<https://devfeed.tech/tags/article.md>), [development](<https://devfeed.tech/tags/development.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [eval](<https://devfeed.tech/tags/eval.md>), [evals](<https://devfeed.tech/tags/evals.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [genai](<https://devfeed.tech/tags/genai.md>), [generation](<https://devfeed.tech/tags/generation.md>), [generative-ai](<https://devfeed.tech/tags/generative-ai.md>), [llm](<https://devfeed.tech/tags/llm.md>), [software](<https://devfeed.tech/tags/software.md>), [software-testing](<https://devfeed.tech/tags/software-testing.md>), [testing](<https://devfeed.tech/tags/testing.md>), [tooling](<https://devfeed.tech/tags/tooling.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

This article presents eval-driven development as a core engineering discipline for trustworthy Generative AI products. It explains why evaluating LLM systems is difficult, including non-deterministic outputs, subjective correctness, AI-based evaluation risks, and failures across retrieval, reasoning, tool calls, and generation. It shares foundational evaluation practices and cautions that teams should plan evaluation early and ground success criteria in their data.

### Source excerpt

How Airbnb teams build trustworthy Generative AI products by treating evaluation as a first-class engineering discipline; not an afterthought.Nestled into the lush hillside, this stunning modern retreat features striking natural wood architecture, terraced balconies, and a serene landscape. By: Rohit Girme, Dan Miller, Mia Zhao, Lifan Yang, Clint Kelly Introduction Generative AI breaks a lot of the assumptions that used to hold true for software testing. Unlike traditional software, LLM outputs are non-deterministic, and "correct" is subjective. Because so much judgment is involved, you often need an AI to evaluate an AI, which introduces its own potential failure modes. Making matters more complicated, a single interaction with an LLM can chain retrieval, reasoning, tool calls, and generation, each of which can fail independently. At Airbnb, we build LLM-powered features across our product, with recent launches including review highlights, AI customer support, smart communication features for guests and hosts, and more. Behind the scenes, we also use AI to help us spot trends and understand what's working, guiding where we improve the product next. Each product team may have its own evaluation criteria, process, workflows, etc. However, these are built on top of some common foundations and principles. An infrastructure team provides tooling and best practices, incorporating learnings across domains so that they are shared with everyone building products at Airbnb. In this article, we wanted to share some of these best practices and learnings with the broader engineering community. Please note that the recommendations here are not intended to be prescriptive; there is no one-size-fits all approach when it comes to running evals. 1. Foundation Evaluating LLM-based systems is challenging work, and this should be planned for at the outset. Without a deliberate strategy, three things tend to happen: False confidence: A generic "helpfulness" metric scores well, you ship,

## Personalizing Airbnb search by learning from the guest journey

DevFeed: [Personalizing Airbnb search by learning from the guest journey](<https://devfeed.tech/articles/personalizing-airbnb-search-by-learning-from-the-guest-journey-1219.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/personalizing-airbnb-search-by-learning-from-the-guest-journey-bcefd1915624?source=rss----53c7c27702d5---4>)

Author: Daochen Zha

Published: 2026-07-21T17:01:04Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [recommendation systems](<https://devfeed.tech/topics/recommendation-systems.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [Representation learning](<https://devfeed.tech/topics/representation-learning.md>), [AI-driven information retrieval](<https://devfeed.tech/topics/ai-driven-information-retrieval.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [airbnb](<https://devfeed.tech/tags/airbnb.md>), [behavior](<https://devfeed.tech/tags/behavior.md>), [booking](<https://devfeed.tech/tags/booking.md>), [browsing](<https://devfeed.tech/tags/browsing.md>), [cancellation](<https://devfeed.tech/tags/cancellation.md>), [challenges](<https://devfeed.tech/tags/challenges.md>), [conversion](<https://devfeed.tech/tags/conversion.md>), [engagement](<https://devfeed.tech/tags/engagement.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [event](<https://devfeed.tech/tags/event.md>), [events](<https://devfeed.tech/tags/events.md>), [feature](<https://devfeed.tech/tags/feature.md>), [features](<https://devfeed.tech/tags/features.md>), [full](<https://devfeed.tech/tags/full.md>), [guests](<https://devfeed.tech/tags/guests.md>), [journey](<https://devfeed.tech/tags/journey.md>), [learning](<https://devfeed.tech/tags/learning.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [model](<https://devfeed.tech/tags/model.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [ranking](<https://devfeed.tech/tags/ranking.md>), [recommendation-system](<https://devfeed.tech/tags/recommendation-system.md>), [search](<https://devfeed.tech/tags/search.md>), [technology](<https://devfeed.tech/tags/technology.md>)

### AI overview

Airbnb replaced hand-crafted guest-history aggregates with a Transformer that combines up to 80 long-term events from seven years (such as bookings and reviews) with up to 200 listing views from the past 21 days. To control training cost, it batches searches through one causal encoder pass and reduces padding with length buckets and sparse search calculations; for serving, it computes guest embeddings daily and combines stored embeddings with live queries. Reported offline booking-label NDCG gains were 0.44% from long-term history, 1.48% after adding recent views, and a further 2.3% from a setwise ranker. Online tests reported gains in uncanceled bookings and related metrics, and reuse of the embeddings in promotional email ranking increased clicks by 5.04%. The system's reported results are specific to Airbnb's experiments.

### Source excerpt

How we built a Transformer-based sequence model that encodes years of guest behavior to surface the right listings at the right time. By: Daochen Zha, Chun How Tan, Xin Liu, Bin Xu, Han Zhao, Xiaowei Liu, Jun Shi, Tracy Yu, Hui Gao, Huiji Gao, Liwei He, Michael Kinoti, Stephanie Moyerman, and Sanjeev Katariya Introduction Planning a trip on Airbnb rarely happens in a single session. A guest searching for a place to stay in San Francisco might browse dozens of listings over several days, leaving behind a trail of views. Typically, over a period of years, that same guest will have accumulated many previous bookings, reviews, and the occasional cancellation. Taken together, these events reveal a great deal about what that guest values in a stay. For years, Airbnb's search ranking captured this through hand-crafted features: aggregated statistics such as total past bookings or average listing price. These worked well, but as the feature count grew into the hundreds, the approach became harder to scale and increasingly limited in expressiveness. In this blog post, we describe how we built a sequence modeling system that encodes the full guest journey using a Transformer, learning richer representations of guest preferences to deliver more personalized search results. An example of a guest journey, which is typically long, exploratory, and complex.Challenges Event sequences per guest present three core challenges. First, they are dominated by listing views, which account for the vast majority of all events -- some guests accumulate hundreds of thousands of them -- making raw sequences computationally intractable to model directly. The distribution of event types, with the majority being listing views. Second, unlike social media platforms, which optimize for engagement, Airbnb optimizes for booking conversion. Bookings are rare, compared to events, and deliberate, whereas a listing view could reflect genuine intent or simply idle browsing. Building a model that generalizes

## From weeks to a day: how we made LLM evaluation fast enough to iterate on

DevFeed: [From weeks to a day: how we made LLM evaluation fast enough to iterate on](<https://devfeed.tech/articles/from-weeks-to-a-day-how-we-made-llm-evaluation-fast-enough-to-iterate-on-1217.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/from-weeks-to-a-day-how-we-made-llm-evaluation-fast-enough-to-iterate-on-14e2d35198b4?source=rss----53c7c27702d5---4>)

Author: Baharak Saberidokht

Published: 2026-07-14T17:01:03Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [Integration testing](<https://devfeed.tech/topics/integration-testing.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [LLM observability](<https://devfeed.tech/topics/llm-observability.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [airbnb](<https://devfeed.tech/tags/airbnb.md>), [approach](<https://devfeed.tech/tags/approach.md>), [bleu](<https://devfeed.tech/tags/bleu.md>), [bugs](<https://devfeed.tech/tags/bugs.md>), [challenges](<https://devfeed.tech/tags/challenges.md>), [component](<https://devfeed.tech/tags/component.md>), [data](<https://devfeed.tech/tags/data.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [eval](<https://devfeed.tech/tags/eval.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [evaluations](<https://devfeed.tech/tags/evaluations.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [fast](<https://devfeed.tech/tags/fast.md>), [foundation](<https://devfeed.tech/tags/foundation.md>), [improvements](<https://devfeed.tech/tags/improvements.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [integration](<https://devfeed.tech/tags/integration.md>), [introduction](<https://devfeed.tech/tags/introduction.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-evaluation](<https://devfeed.tech/tags/llm-evaluation.md>), [technology](<https://devfeed.tech/tags/technology.md>)

### AI overview

Airbnb describes a four-layer infrastructure approach that reduced LLM evaluation and fix cycles to same-day iteration. First, distinguish data uncertainty from judge uncertainty, and test whether conclusions persist across judges, metrics, and sample groupings. Second, cache generated references by sample and configuration, and judge scores by sample, output, judge configuration, and metric; this makes repeated evaluations reproducible, cheaper, and resumable. Third, apply targeted rank-under-50 LoRA micro adapters that train in under an hour, with expert-domain regression checks and human review for high-uncertainty outputs, plus canary deployment and rollback. The authors caution that stacked patches can interfere, so overlapping patches should be fused, accumulated patches eventually retrained, and unused patches unloaded. Finally, run a small representative set through the full production path, combining traffic-weighted sampling with extra coverage of locales, modalities, and prior incident cases, and measure quality and tail latency. The layers depend on one another: deterministic measurement enables fast fixes and meaningful end-to-end checks.

### Source excerpt

Training an LLM is the easy part. The hard part is designing experiments and evaluations that you can trust enough to know whether the new model is actually an improvement. By: Baharak Saberidokht Introduction Shipping a production LLM system means iterating fast on improvements to something that is, by construction, non-deterministic. Models drift, judges disagree with themselves, references regenerate as different strings, and bugs may persist until the next release, because retraining takes weeks. Most of this friction comes from infrastructure challenges, not model quality, and the fixes come from classical software engineering techniques. At Airbnb, we built reliable LLM infrastructure by addressing four layers. Three correspond to engineering enhancements we've made; the fourth is the integration layer that ties the rest together -- the one that is easiest to overlook, because each individual component looks fine in isolation. The approach rests on two observations: the seams are where things break, and finding those breaks requires exercising the full path, not just validating each component in isolation. Figure 1. The four layers of the production LLM stack. Bounded model mutation requires trustworthy measurement, and end-to-end validation requires the eval foundation to be fast enough to run on the combined path. Layer 1: Name it before trying to remove it Layer 1 is diagnostic framing of evaluation noise. This layer addresses two different sources of indeterminacy: data and judging uncertainty. Classical ML metrics are deterministic: F1, BLEU, and accuracy return the same number on the same input. With LLMs in the evaluation loop, that assumption dies. Judges score identical inputs differently across runs, and LLM-generated references regenerate as different strings. A two percent score movement can mean the model improved, the judge drifted, the references shifted, or some combination. We cannot tell which without naming which kind of noise we are looking

## Scaling beyond one: How Airbnb evolved its data architecture for a multi-product world

DevFeed: [Scaling beyond one: How Airbnb evolved its data architecture for a multi-product world](<https://devfeed.tech/articles/scaling-beyond-one-how-airbnb-evolved-its-data-architecture-for-a-multi-product-world-1222.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/scaling-beyond-one-how-airbnb-evolved-its-data-architecture-for-a-multi-product-world-6125645d470c?source=rss----53c7c27702d5---4>)

Author: Patrick Lam

Published: 2026-06-09T17:01:02Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [data-architecture](<https://devfeed.tech/topics/data-architecture.md>), [Data Infrastructure](<https://devfeed.tech/topics/data-infrastructure.md>), [data-modeling](<https://devfeed.tech/topics/data-modeling.md>), [data](<https://devfeed.tech/topics/data.md>)

Tags: [analytics](<https://devfeed.tech/tags/analytics.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [data](<https://devfeed.tech/tags/data.md>), [data-architecture](<https://devfeed.tech/tags/data-architecture.md>), [data-engineering](<https://devfeed.tech/tags/data-engineering.md>), [data-infrastructure](<https://devfeed.tech/tags/data-infrastructure.md>), [data-modeling](<https://devfeed.tech/tags/data-modeling.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [offline](<https://devfeed.tech/tags/offline.md>), [post](<https://devfeed.tech/tags/post.md>), [technology](<https://devfeed.tech/tags/technology.md>)

### AI overview

Airbnb's data and analytics engineering teams evolved a decade-old offline data warehouse to support Homes, Experiences, and Services. The article examines the trade-offs between separate product-specific data models and a unified monolithic model while describing the need for a consistent, flexible, and scalable data foundation.

### Source excerpt

How Airbnb's data engineers and analytics engineers built a consistent and flexible data modeling framework to support the expansion into Homes, Experiences, and Services. By: Patrick Lam, Namrata Lamba, Jamie Stober With the May 2025 Summer Release, Airbnb redesigned its app, relaunched Experiences, and debuted Services, pushing us beyond our traditional Homes focus. For the data teams, this meant rapidly evolving a decade-old infrastructure to integrate two brand-new product pillars. Our data engineers and analytics engineers rose to the challenge by building a consistent and flexible framework to serve as a robust and scalable data foundation for the next decade of growth. But getting there wasn't straightforward. This fundamental shift surfaced a critical question for our data organization: How do you evolve your offline data architecture to support new product lines without introducing disorder in vital analytics services? We knew the approach we took would have long-lasting implications. A fragmented strategy risked creating data silos, inconsistent analytics, and a tangled web of technical debt that would likely slow down future innovation. In this post, we'll take you behind the scenes to share key decisions that we made, the framework that emerged, and the lessons that helped reshape our offline data warehouse for the future. Note that we focus specifically on our offline data warehouse (the analytics-oriented data infrastructure owned by our data engineers and analytics engineers) rather than the online data systems that serve the app directly, as the two domains have fundamentally different requirements, constraints, and design philosophies that warrant separate treatment. The core dilemma: separate vs. monolithic The first and most critical question was how to structure offline data for the new, three-product world, with Homes, a refreshed Experiences product, and the new Services offering. This involved a trade-off between two main approaches: Separate

## Sitar-agent: Building a reliable dynamic configuration sidecar at scale

DevFeed: [Sitar-agent: Building a reliable dynamic configuration sidecar at scale](<https://devfeed.tech/articles/sitar-agent-building-a-reliable-dynamic-configuration-sidecar-at-scale-1223.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/sitar-agent-building-a-reliable-dynamic-configuration-sidecar-at-scale-b7e00c152068?source=rss----53c7c27702d5---4>)

Author: Bo Teng

Published: 2026-06-04T17:01:04Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [configuration](<https://devfeed.tech/topics/configuration.md>), [Kubernetes](<https://devfeed.tech/topics/kubernetes.md>), [Amazon S3](<https://devfeed.tech/topics/amazon-s3.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [distributed-systems](<https://devfeed.tech/tags/distributed-systems.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [s3](<https://devfeed.tech/tags/s3.md>), [scale](<https://devfeed.tech/tags/scale.md>), [software-architecture](<https://devfeed.tech/tags/software-architecture.md>), [software-development](<https://devfeed.tech/tags/software-development.md>), [sync](<https://devfeed.tech/tags/sync.md>)

### AI overview

Airbnb describes Sitar-agent, a Kubernetes sidecar that reliably delivers dynamic configuration to service pods. It bootstraps from S3 snapshots, then synchronizes newer configuration from the Sitar Service.

### Source excerpt

How Airbnb built a Kubernetes sidecar to deliver dynamic configuration reliably at scale. By: Bo Teng, Cosmo Qiu, Siyuan Zhou, Ankur Soni, Xin Huang, Willis Harvey Introduction In our previous post, we explored Airbnb's dynamic configuration system, Sitar, with a focus on service architecture and configuration change safety. Now for the harder question: once a config change is committed, which happens several times each minute, how does it actually reach the thousands of Airbnb's service instances reliably, quickly, and without redeploying the services? This post describes sitar agent: a lightweight Kubernetes sidecar that runs alongside every subscribed service pod, continuously synchronizing the latest configurations from the service backend and making them available on the local filesystem for reads. In this post, we will first go through the configuration delivery life cycle, and then discuss some key design choices for the sitar-agent sidecar. Config delivery life cycle The diagram below illustrates the end-to-end journey of a configuration change, from the developer-facing layer to the production service fleet. Sitar config delivery lifecycle Step 1 -- Config creation/update Developers create or update configuration values through either Git flow or the web UI. These changes are committed to the Sitar Service, where they are stored with full versioning, change logs, and ACL enforcement. Step 2 -- Hourly snapshot upload The Snapshot Service periodically packages the full state of all config groups and uploads compressed snapshots to AWS S3. Step 3.1 -- Preload snapshot from S3 (on pod startup) When a production service pod starts, the sitar-agent sidecar runs first. It downloads the latest snapshot for each subscribed tenant's configs from S3 to the mounted disk (shared between sitar-agent and the main container). This allows the agent to bootstrap from a known-good state without fetching every config from the Sitar Service from scratch on every restart. Preloadin

## When history fails you, borrow from geography

DevFeed: [When history fails you, borrow from geography](<https://devfeed.tech/articles/when-history-fails-you-borrow-from-geography-1224.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/when-history-fails-you-borrow-from-geography-915a72b91b5c?source=rss----53c7c27702d5---4>)

Author: Harrison Katz

Published: 2026-06-02T17:01:04Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [Predictive AI](<https://devfeed.tech/topics/predictive-ai.md>), [Time Series](<https://devfeed.tech/topics/time-series.md>), [dataset](<https://devfeed.tech/topics/dataset.md>)

Tags: [2020](<https://devfeed.tech/tags/2020.md>), [2022](<https://devfeed.tech/tags/2022.md>), [ai](<https://devfeed.tech/tags/ai.md>), [airbnb](<https://devfeed.tech/tags/airbnb.md>), [approach](<https://devfeed.tech/tags/approach.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [bayesian](<https://devfeed.tech/tags/bayesian.md>), [border](<https://devfeed.tech/tags/border.md>), [changes](<https://devfeed.tech/tags/changes.md>), [covid](<https://devfeed.tech/tags/covid.md>), [data](<https://devfeed.tech/tags/data.md>), [data-modeling](<https://devfeed.tech/tags/data-modeling.md>), [data-science](<https://devfeed.tech/tags/data-science.md>), [demand](<https://devfeed.tech/tags/demand.md>), [forecasting](<https://devfeed.tech/tags/forecasting.md>), [geography](<https://devfeed.tech/tags/geography.md>), [history](<https://devfeed.tech/tags/history.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [model](<https://devfeed.tech/tags/model.md>), [recovery](<https://devfeed.tech/tags/recovery.md>), [sequence](<https://devfeed.tech/tags/sequence.md>), [standard](<https://devfeed.tech/tags/standard.md>), [technology](<https://devfeed.tech/tags/technology.md>), [time](<https://devfeed.tech/tags/time.md>), [travel](<https://devfeed.tech/tags/travel.md>)

### AI overview

Airbnb addressed sparse local data during COVID's staggered recovery by using earlier-recovering travel corridors as informative priors for structurally similar corridors that reopened later. In a hierarchical Bayesian framework, signals such as booking lead times were weighted by similarities in traveler mix, domestic versus international demand, and accommodation mix; as local observations accumulated, they increasingly shaped each corridor's estimates. This enabled timely forecasts during reopenings, reclosures, and uneven vaccine rollouts. The method depends on broad geographic coverage, comparable data across markets, and a model that can incorporate and update corridor-level priors. The text reports operational usefulness but gives no quantified forecast-accuracy results, and notes the approach is not suitable for every problem.

### Source excerpt

How Airbnb used sequential geographic recovery signals and prior propagation to generate reliable corridor-level forecasts when local data was scarce. By: Harrison Katz The problem with unprecedented shocks Almost every forecasting system is built on the same implicit assumption: the future will resemble the past. You train on historical data, you validate on holdout periods, and you trust that past patterns will at least roughly indicate future performance. When this assumption breaks, the model does not gracefully degrade; it fails confidently. It produces precise, well-calibrated intervals around the wrong answer. The acute phase of COVID, from early to late 2020, was a clear illustration of this, and we wrote about it in a previous post. But the more interesting forecasting problem was not the shutdown. It was everything that came after. The period from late 2020 through 2022 was not a single coherent regime. It was a sequence of overlapping, asynchronous changes: vaccine rollouts that reached some markets months before others, border reopenings that followed their own country-level timelines, reclosures triggered by new variants that hit different corridors (a pairing of the traveler's origin city and destination city) at different moments. Demand was not recovering uniformly. It was rebounding unevenly across every corner of the world, in ways that had no historical precedent and no single governing pattern. The standard response to a shock is to wait for each affected market to accumulate its own post-shock data and retrain locally. But Covid was among the biggest shocks the travel industry has faced in decades. With markets worldwide reopening and reclosing on staggered schedules, waiting for markets to settle meant forecasting blind for months at a time, across all markets, just when timely projections were most needed, in the circumstances. So we started building something different. When we could not simply look backward in time for relevant examples, we

## Scaling Airbnb's identity graph with a unified knowledge graph infrastructure

DevFeed: [Scaling Airbnb's identity graph with a unified knowledge graph infrastructure](<https://devfeed.tech/articles/scaling-airbnb-s-identity-graph-with-a-unified-knowledge-graph-infrastructure-1221.md>)

Original publisher: [Read original article](<https://medium.com/airbnb-engineering/scaling-airbnbs-identity-graph-with-a-unified-knowledge-graph-infrastructure-ebac467b7836?source=rss----53c7c27702d5---4>)

Author: Lucen Zhao

Published: 2026-05-19T17:01:01Z

Content type: article

Language: en

Sources: [Airbnb](<https://devfeed.tech/sources/the-airbnb-tech-blog-medium.md>)

Topics: [graph-database](<https://devfeed.tech/topics/graph-database.md>), [Graphs](<https://devfeed.tech/topics/graphs.md>), [software-architecture](<https://devfeed.tech/topics/software-architecture.md>), [Cloud Architecture](<https://devfeed.tech/topics/cloud-architecture.md>), [Cache](<https://devfeed.tech/topics/cache.md>), [Latency](<https://devfeed.tech/topics/latency.md>), [Managed Services](<https://devfeed.tech/topics/managed-services.md>)

Tags: [aggregation](<https://devfeed.tech/tags/aggregation.md>), [airbnb](<https://devfeed.tech/tags/airbnb.md>), [applications](<https://devfeed.tech/tags/applications.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [ashish-jain](<https://devfeed.tech/tags/ashish-jain.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [caching](<https://devfeed.tech/tags/caching.md>), [complexity](<https://devfeed.tech/tags/complexity.md>), [data](<https://devfeed.tech/tags/data.md>), [data-platform](<https://devfeed.tech/tags/data-platform.md>), [data-products](<https://devfeed.tech/tags/data-products.md>), [data-storage](<https://devfeed.tech/tags/data-storage.md>), [database](<https://devfeed.tech/tags/database.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [events](<https://devfeed.tech/tags/events.md>), [evolution](<https://devfeed.tech/tags/evolution.md>), [graph](<https://devfeed.tech/tags/graph.md>), [graph-database](<https://devfeed.tech/tags/graph-database.md>), [identity](<https://devfeed.tech/tags/identity.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [knowledge-graph](<https://devfeed.tech/tags/knowledge-graph.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [technology](<https://devfeed.tech/tags/technology.md>)

### AI overview

Airbnb migrated its identity graph from a third-party graph service to an internally managed platform built on JanusGraph, DynamoDB, and OpenSearch. The graph contains 7 billion nodes and 11 billion edges, adds about 5 million edges daily, and serves queries that commonly traverse 4-8 hops. The platform provides tenant isolation, schema and index management, and Thrift APIs. To reduce latency and operational issues, Airbnb added custom transactions using DynamoDB conditional writes and transactions, parallelized data retrieval for high-fanout queries, and improved tracing. Client-side Gremlin rewrites also removed costly Path steps where possible and simplified side-effect aggregations. Airbnb reports better query performance and stability than the prior vendor, no longer needing periodic manual reboots, and write throughput scaled to ten times the previous solution in load tests. These results are reported by Airbnb; the text gives no absolute latency figures.

### Source excerpt

How Airbnb shifts from PaaS to an internal knowledge graph infrastructure at scale. By: Lucen Zhao, Shukun Yang, Ashish Jain Knowledge graphs offer a natural and powerful way to represent relationships between entities. Many real-world systems are fundamentally about connections. Airbnb's identity graph captures relationships between users in a graph database. The identity graph serves aggregated insights that enable user identity resolution and relationship understanding. These capabilities support a wide range of Trust and Safety use cases, from detecting suspicious activities to identifying linked accounts. Over time, the identity graph has grown into one of the largest and most complex graph data products at Airbnb, both in terms of scale and the complexity of queries it supports. In 2024, Airbnb began investing in a new, internally managed, paved-path graph data platform to build a unified knowledge graph infrastructure. Airbnb's identity graph became one of the first systems to adopt this platform. In this post, we'll walk through the foundations and challenges of the identity graph, introduce the architecture behind the graph infrastructure, and highlight several key optimizations that emerged during the onboarding process. Airbnb's identity graph Airbnb's identity graph is a critical foundation layer, playing an important role in Trust and Safety applications. It contains two major components: Graph data storage: a storage layer composed of a graph database and a key-value (KV) caching layer. It models users and relationships as vertices and edges. Most data is ingested in near real-time through asynchronous events and served through low-latency, real-time service calls. Graph service: this service provides a unified interface for accessing graph data. It retrieves data from underlying sources, including the graph database, applies aggregation logic or models as needed, and serves the results to downstream customer services. Evolution of the identity graph a