# Spotify

Spotify's official technology blog

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

## AI Product Management at Spotify: A View From Three Angles

DevFeed: [AI Product Management at Spotify: A View From Three Angles](<https://devfeed.tech/articles/ai-product-management-at-spotify-a-view-from-three-angles-126492.md>)

Original publisher: [Read original article](<https://engineering.atspotify.com/2026/11/ai-product-management-at-spotify-a-view-from-three-angles/>)

Author: Spotify Engineering

Published: 2026-11-09T17:53:35Z

Content type: article

Language: en

Sources: [Spotify](<https://devfeed.tech/sources/spotify-engineering-blog.md>), [Spotify Engineering](<https://devfeed.tech/sources/spotify-engineering.md>)

Topics: [Product Management](<https://devfeed.tech/topics/product-management.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Spotify](<https://devfeed.tech/topics/spotify.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-product-management](<https://devfeed.tech/tags/ai-product-management.md>), [product](<https://devfeed.tech/tags/product.md>), [spotify](<https://devfeed.tech/tags/spotify.md>)

### AI overview

The article describes three connected AI product management roles at Spotify, each with a different customer and measure of impact. In Tech Research, the PM identifies questions spanning teams, tests ideas against production needs such as evaluation, latency, cost, data, and integration, and connects evidence to organizational decisions while research feedback loops are long. For a recommendation foundation model, the PM shapes learned behavior through training data and evaluation, compares offline results with online impact, and must show partner teams that the shared model meets quality, latency, and cost needs before they adopt it. For the user experience, the PM turns probabilistic capabilities into a useful, trustworthy product for open-ended requests, builds datasets and tailored evaluators because standard A/B and offline tests do not cover every interaction, and routes conversational user feedback back to model and research teams. The roles share product fundamentals but differ in what they own and how success is demonstrated.

### Source excerpt

Almost every article or podcast about product management and AI refers to "the AI PM" as one job. But what... The post AI Product Management at Spotify: A View From Three Angles appeared first on Spotify Engineering.

## Advancing Generative Recommendation Inference: our collaboration with NVIDIA on TensorRT LLM

DevFeed: [Advancing Generative Recommendation Inference: our collaboration with NVIDIA on TensorRT LLM](<https://devfeed.tech/articles/advancing-generative-recommendation-inference-our-collaboration-with-nvidia-on-tensorrt-llm-88452.md>)

Original publisher: [Read original article](<https://engineering.atspotify.com/2026/10/advancing-generative-recommendation-inference-our-collaboration-with-nvidia-on-tensorrt-llm/>)

Author: Spotify Engineering

Published: 2026-10-16T14:29:18Z

Content type: article

Language: en

Sources: [Spotify](<https://devfeed.tech/sources/spotify-engineering-blog.md>), [Spotify Engineering](<https://devfeed.tech/sources/spotify-engineering.md>)

Topics: [recommendation systems](<https://devfeed.tech/topics/recommendation-systems.md>), [TensorRT-LLM](<https://devfeed.tech/topics/tensorrt-llm.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [personalization](<https://devfeed.tech/topics/personalization.md>), [ab-mcts](<https://devfeed.tech/topics/ab-mcts.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [vllm](<https://devfeed.tech/topics/vllm.md>)

Tags: [batching](<https://devfeed.tech/tags/batching.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [collaboration](<https://devfeed.tech/tags/collaboration.md>), [cost-efficiency](<https://devfeed.tech/tags/cost-efficiency.md>), [costs](<https://devfeed.tech/tags/costs.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [deployments](<https://devfeed.tech/tags/deployments.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [embeddings](<https://devfeed.tech/tags/embeddings.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [generative](<https://devfeed.tech/tags/generative.md>), [inference](<https://devfeed.tech/tags/inference.md>), [llm](<https://devfeed.tech/tags/llm.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [recommendation](<https://devfeed.tech/tags/recommendation.md>), [tensorrt](<https://devfeed.tech/tags/tensorrt.md>), [tensorrt-llm](<https://devfeed.tech/tags/tensorrt-llm.md>)

### AI overview

Spotify and NVIDIA collaborated to extend TensorRT-LLM with beam search, multi-item scoring, and soft-token support for generative recommendation workloads. The post describes inference engineering tradeoffs and reports cost reductions of up to 54% for beam-search workloads and 7% for scoring workloads on a 1.7B Qwen3 variant.

### Source excerpt

At Spotify, every surface and interaction is personalized to you as a user: the content on the Home page, the... The post Advancing Generative Recommendation Inference: our collaboration with NVIDIA on TensorRT LLM appeared first on Spotify Engineering.

## Introducing: Spotify Technology. Proven at Spotify, now yours.

DevFeed: [Introducing: Spotify Technology. Proven at Spotify, now yours.](<https://devfeed.tech/articles/introducing-spotify-technology-proven-at-spotify-now-yours-77056.md>)

Original publisher: [Read original article](<https://engineering.atspotify.com/2026/10/introducing-spotify-technology-proven-at-spotify-now-yours/>)

Author: Spotify Engineering

Published: 2026-10-08T12:30:10Z

Content type: release

Language: en

Sources: [Spotify](<https://devfeed.tech/sources/spotify-engineering-blog.md>), [Spotify Engineering](<https://devfeed.tech/sources/spotify-engineering.md>)

Topics: [Spotify](<https://devfeed.tech/topics/spotify.md>), [AI Trends](<https://devfeed.tech/topics/ai-trends.md>), [engineering-leadership](<https://devfeed.tech/topics/engineering-leadership.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agent-readiness](<https://devfeed.tech/tags/agent-readiness.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [ai](<https://devfeed.tech/tags/ai.md>), [alignment](<https://devfeed.tech/tags/alignment.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [artifacts](<https://devfeed.tech/tags/artifacts.md>), [autonomy](<https://devfeed.tech/tags/autonomy.md>), [backend](<https://devfeed.tech/tags/backend.md>), [backstage](<https://devfeed.tech/tags/backstage.md>), [books](<https://devfeed.tech/tags/books.md>), [build](<https://devfeed.tech/tags/build.md>), [building](<https://devfeed.tech/tags/building.md>), [builds](<https://devfeed.tech/tags/builds.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [connect](<https://devfeed.tech/tags/connect.md>), [context](<https://devfeed.tech/tags/context.md>), [developer-experience](<https://devfeed.tech/tags/developer-experience.md>), [developer-tools](<https://devfeed.tech/tags/developer-tools.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [product](<https://devfeed.tech/tags/product.md>), [spotify](<https://devfeed.tech/tags/spotify.md>), [technology](<https://devfeed.tech/tags/technology.md>)

### AI overview

Spotify Technology brings Spotify's developer products under one brand as a modular system for the product development lifecycle. Portal by Spotify provides organizational context; Xirp carries that context into work across coding agents; Confidence turns experimentation into product intelligence to inform what teams build next. The post says the products share a context layer and agent runtime, and can be adopted starting with an individual engineer, a team, or a full R&D organization. Spotify says it developed these tools to balance team autonomy with alignment, tested them on its own production workloads, and offers them externally as open-source or commercial products. It frames AI-driven increases in software output as intensifying fragmentation, operational costs, and difficulty measuring results, and says it is deepening product connections and investing in agent readiness. Its longer-term aim is to provide connective infrastructure that makes organizational context and learning available to people and agents.

### Source excerpt

Introducing Spotify Technology. The post Introducing: Spotify Technology. Proven at Spotify, now yours. appeared first on Spotify Engineering.

## AI Changed How Spotify Builds. What We Learned (and Fixed) About Quality at Higher Velocity

DevFeed: [AI Changed How Spotify Builds. What We Learned (and Fixed) About Quality at Higher Velocity](<https://devfeed.tech/articles/ai-changed-how-spotify-builds-what-we-learned-and-fixed-about-quality-at-higher-velocity-41282.md>)

Original publisher: [Read original article](<https://engineering.atspotify.com/2026/9/ai-changed-how-spotify-builds-what-we-learned-and-fixed-about-quality-at-higher-velocity/>)

Author: Spotify Engineering

Published: 2026-09-16T19:13:53Z

Content type: article

Language: en

Sources: [Spotify](<https://devfeed.tech/sources/spotify-engineering-blog.md>), [Spotify Engineering](<https://devfeed.tech/sources/spotify-engineering.md>)

Topics: [Spotify](<https://devfeed.tech/topics/spotify.md>), [Production Engineering](<https://devfeed.tech/topics/production-engineering.md>), [Code quality](<https://devfeed.tech/topics/code-quality.md>), [reliability-engineering](<https://devfeed.tech/topics/reliability-engineering.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [Post Mortem](<https://devfeed.tech/topics/post-mortem.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [engineering-culture](<https://devfeed.tech/topics/engineering-culture.md>), [scheduling](<https://devfeed.tech/topics/scheduling.md>), [mobile-app-performance](<https://devfeed.tech/topics/mobile-app-performance.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-assisted](<https://devfeed.tech/tags/ai-assisted.md>), [ai-assisted-development](<https://devfeed.tech/tags/ai-assisted-development.md>), [ai-authored](<https://devfeed.tech/tags/ai-authored.md>), [ai-slop](<https://devfeed.tech/tags/ai-slop.md>), [alerting](<https://devfeed.tech/tags/alerting.md>), [app](<https://devfeed.tech/tags/app.md>), [app-quality](<https://devfeed.tech/tags/app-quality.md>), [automated](<https://devfeed.tech/tags/automated.md>), [automation](<https://devfeed.tech/tags/automation.md>), [backend](<https://devfeed.tech/tags/backend.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [capacity-planning](<https://devfeed.tech/tags/capacity-planning.md>), [catalog](<https://devfeed.tech/tags/catalog.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [quality](<https://devfeed.tech/tags/quality.md>), [spotify](<https://devfeed.tech/tags/spotify.md>)

### AI overview

Spotify says AI-assisted development increased the pace and volume of change, exposing gaps in verification and capacity rather than producing a distinct pattern of failures from AI-authored code. During the incidents reviewed so far, the company found no material direct contribution from AI-authored code, but saw change volume outpace some review, testing, rollout, and observability controls. It has improved content pipeline alerting and scheduling, prioritized new uploads, strengthened safeguards and rollback for automated fleet changes, and expanded app quality signals to catch gradual regressions. Industry-wide AI demand has also made compute less predictable, complicating regional failovers; Spotify doubled reserved edge capacity and is developing more gradual traffic shifting. Merged changes more than doubled year over year in August, while the share of quality and optimization work rose from 27% to 31%; Spotify reports no corresponding rise in its rework metric, but is monitoring increasing code complexity and PR size without changing its thresholds.

### Source excerpt

Quality and reliability have always been a point of pride for Spotify. We run an extraordinarily complex... The post AI Changed How Spotify Builds. What We Learned (and Fixed) About Quality at Higher Velocity appeared first on Spotify Engineering.

## Why Spotify Is Not Using Bayesian A/B Testing

DevFeed: [Why Spotify Is Not Using Bayesian A/B Testing](<https://devfeed.tech/articles/why-spotify-is-not-using-bayesian-a-b-testing-156.md>)

Original publisher: [Read original article](<https://engineering.atspotify.com/2026/9/why-spotify-is-not-using-bayesian-a-b-testing/>)

Author: Spotify Engineering

Published: 2026-09-08T13:18:44Z

Content type: article

Language: en

Sources: [Spotify](<https://devfeed.tech/sources/spotify-engineering-blog.md>), [Spotify Engineering](<https://devfeed.tech/sources/spotify-engineering.md>)

Topics: [A/B Testing](<https://devfeed.tech/topics/a-b-testing.md>), [product-experimentation](<https://devfeed.tech/topics/product-experimentation.md>), [sequential-testing](<https://devfeed.tech/topics/sequential-testing.md>), [Developer Tools](<https://devfeed.tech/topics/developer-tools.md>), [Data science and engineering](<https://devfeed.tech/topics/data-science-and-engineering.md>), [Spotify](<https://devfeed.tech/topics/spotify.md>)

Tags: [a-b-testing](<https://devfeed.tech/tags/a-b-testing.md>), [bayesian](<https://devfeed.tech/tags/bayesian.md>), [bayesian-inference](<https://devfeed.tech/tags/bayesian-inference.md>), [data](<https://devfeed.tech/tags/data.md>), [data-science](<https://devfeed.tech/tags/data-science.md>), [spotify](<https://devfeed.tech/tags/spotify.md>), [testing](<https://devfeed.tech/tags/testing.md>)

### AI overview

Spotify says it has not added Bayesian inference because supporting two modes would add planning, monitoring, interpretation, and configuration costs without enough benefit for its experimentation goals. The post argues Bayesian A/B testing is a family of setups whose guarantees depend on the prior, likelihood, and stopping rule. Flat-prior posterior thresholds can reproduce frequentist peeking behavior; Bayes-factor stopping can control false positives under peeking, and a well-calibrated empirical Bayes prior can shrink winner's-curse estimates and control false discovery rates across metrics. Those benefits require representative historical data and careful calibration; pooling unlike programs or metrics, winner-biased archives, or drift can undermine results. The authors describe many Bayesian and frequentist procedures as equivalent or closely related, while noting decision rules can differ for some cost functions. They recommend choosing required guarantees based on program goals; Spotify judges prior maintenance and the costs of a second framework outweigh its benefits today.

### Source excerpt

Clearing the confusion about what Bayesian A/B testing is. The post Why Spotify Is Not Using Bayesian A/B Testing appeared first on Spotify Engineering.

## Portal by Spotify cut my Claude Code token usage by 90%

DevFeed: [Portal by Spotify cut my Claude Code token usage by 90%](<https://devfeed.tech/articles/portal-by-spotify-cut-my-claude-code-token-usage-by-90-155.md>)

Original publisher: [Read original article](<https://engineering.atspotify.com/2026/9/portal-by-spotify-cut-my-claude-code-token-usage-by-90/>)

Author: Spotify Engineering

Published: 2026-09-03T16:06:40Z

Content type: article

Language: en

Sources: [Spotify](<https://devfeed.tech/sources/spotify-engineering-blog.md>), [Spotify Engineering](<https://devfeed.tech/sources/spotify-engineering.md>)

Topics: [AI Bots](<https://devfeed.tech/topics/ai-bots.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-coding](<https://devfeed.tech/tags/ai-coding.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [cli](<https://devfeed.tech/tags/cli.md>), [gemini](<https://devfeed.tech/tags/gemini.md>), [mcp](<https://devfeed.tech/tags/mcp.md>)

### AI overview

The article describes using Portal modes to route repetitive AI coding work, such as reading large files and generating predictable code, to a cheaper model while reserving Claude for harder problems.

### Source excerpt

Most of what an AI coding agent does for me isn't thinking. It's I/O. The post Portal by Spotify cut my Claude Code token usage by 90% appeared first on Spotify Engineering.

## When Can LLMs Replace Humans in A/B Tests?

DevFeed: [When Can LLMs Replace Humans in A/B Tests?](<https://devfeed.tech/articles/when-can-llms-replace-humans-in-a-b-tests-154.md>)

Original publisher: [Read original article](<https://engineering.atspotify.com/2026/8/when-can-llms-replace-humans-in-a-b-tests/>)

Author: Spotify Engineering

Published: 2026-08-13T18:57:22Z

Content type: article

Language: en

Sources: [Spotify](<https://devfeed.tech/sources/spotify-engineering-blog.md>), [Spotify Engineering](<https://devfeed.tech/sources/spotify-engineering.md>)

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [dataset](<https://devfeed.tech/topics/dataset.md>), [experiments](<https://devfeed.tech/topics/experiments.md>), [Algorithms](<https://devfeed.tech/topics/algorithms.md>)

Tags: [a-b-testing](<https://devfeed.tech/tags/a-b-testing.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [archive](<https://devfeed.tech/tags/archive.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [bias](<https://devfeed.tech/tags/bias.md>), [clinical-trials](<https://devfeed.tech/tags/clinical-trials.md>), [data](<https://devfeed.tech/tags/data.md>), [data-science](<https://devfeed.tech/tags/data-science.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [error](<https://devfeed.tech/tags/error.md>), [estimator](<https://devfeed.tech/tags/estimator.md>), [fine-tuning](<https://devfeed.tech/tags/fine-tuning.md>), [gradient-boosted-trees](<https://devfeed.tech/tags/gradient-boosted-trees.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llms](<https://devfeed.tech/tags/llms.md>), [predictions](<https://devfeed.tech/tags/predictions.md>), [tests](<https://devfeed.tech/tags/tests.md>)

### AI overview

LLM predictions can recover human treatment effects only under unverified assumptions of surrogacy and comparability, so they cannot reliably replace user experiments for new interventions. In Upworthy headline tests, raw gpt-4o-mini predictions recovered 39% of the observed effect; OLS calibration failed a held-out falsification test, while random forests and gradient-boosted trees produced estimates within sampling error. Averaging multiple LLM outputs per unit can reduce sampling noise. Calibration requires actual user data, may become invalid as treatments or models change, and is most credible when new tests resemble historical ones. The excerpt suggests using predictions to screen weak ideas or as covariates for variance reduction, while retaining human experiments for validation and innovation.

### Source excerpt

TL;DR: LLM predictions can stand in for human outcomes in A/B tests, but only by assumption, not by design.... The post When Can LLMs Replace Humans in A/B Tests? appeared first on Spotify Engineering.

## Indexing the Data Lake for Online Point Queries

DevFeed: [Indexing the Data Lake for Online Point Queries](<https://devfeed.tech/articles/indexing-the-data-lake-for-online-point-queries-152.md>)

Original publisher: [Read original article](<https://engineering.atspotify.com/2026/7/indexing-the-data-lake-for-online-point-queries/>)

Author: Spotify Engineering

Published: 2026-07-27T20:34:28Z

Content type: article

Language: en

Sources: [Spotify](<https://devfeed.tech/sources/spotify-engineering-blog.md>), [Spotify Engineering](<https://devfeed.tech/sources/spotify-engineering.md>)

Topics: [data lake](<https://devfeed.tech/topics/data-lake.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [data-engineering](<https://devfeed.tech/topics/data-engineering.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [olap](<https://devfeed.tech/topics/olap.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [aggregate](<https://devfeed.tech/tags/aggregate.md>), [aggregates](<https://devfeed.tech/tags/aggregates.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [alignment](<https://devfeed.tech/tags/alignment.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [applications](<https://devfeed.tech/tags/applications.md>), [data](<https://devfeed.tech/tags/data.md>), [data-lake](<https://devfeed.tech/tags/data-lake.md>), [indexing](<https://devfeed.tech/tags/indexing.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [latency](<https://devfeed.tech/tags/latency.md>), [low-latency](<https://devfeed.tech/tags/low-latency.md>), [online](<https://devfeed.tech/tags/online.md>), [point](<https://devfeed.tech/tags/point.md>), [queries](<https://devfeed.tech/tags/queries.md>), [services](<https://devfeed.tech/tags/services.md>), [spotify](<https://devfeed.tech/tags/spotify.md>)

### AI overview

Random Access Parquet (RAP) enables low-latency point lookups over existing Parquet files by using an external index to map keys to exact files and rows, then fetching needed data with parallel ranged reads. This avoids the dependent footer, metadata, and column scans that make distributed SQL engines such as Trino and BigQuery slow for single-row queries. RAP's index is definitive, unlike Parquet Bloom filters and PageIndex, which only narrow scans. Optional file preparation can further reduce bytes or requests: sort or bucket by key, group values per key, use one page per key or independently addressable ZSTD frames, combine queried fields into blobs, or interleave columns. These choices trade off against batch-query pruning, compression, PageIndex size, or extra I/O for conventional readers. Small values and aggregates can also be hoisted into the index to avoid storage reads. The approach serves online and analytical workloads from the same lake files, though the text describes optimizations and potential benefits rather than measured performance results.

### Source excerpt

Companies like Spotify need vast quantities of data accessible at low latency for online services and,... The post Indexing the Data Lake for Online Point Queries appeared first on Spotify Engineering.

## Content Ingestion & Podcast Video Incident Report

DevFeed: [Content Ingestion & Podcast Video Incident Report](<https://devfeed.tech/articles/content-ingestion-podcast-video-incident-report-149.md>)

Original publisher: [Read original article](<https://engineering.atspotify.com/2026/7/content-ingestion-and-podcast-video-incident-report/>)

Author: Spotify Engineering

Published: 2026-07-20T16:24:48Z

Content type: news

Language: en

Sources: [Spotify](<https://devfeed.tech/sources/spotify-engineering-blog.md>), [Spotify Engineering](<https://devfeed.tech/sources/spotify-engineering.md>)

Topics: [Post Mortem](<https://devfeed.tech/topics/post-mortem.md>), [podcasting](<https://devfeed.tech/topics/podcasting.md>), [Production Engineering](<https://devfeed.tech/topics/production-engineering.md>), [transcoder](<https://devfeed.tech/topics/transcoder.md>), [incident management](<https://devfeed.tech/topics/incident-management.md>), [performance-engineering](<https://devfeed.tech/topics/performance-engineering.md>), [Spotify](<https://devfeed.tech/topics/spotify.md>)

Tags: [alert](<https://devfeed.tech/tags/alert.md>), [alerts](<https://devfeed.tech/tags/alerts.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [apps](<https://devfeed.tech/tags/apps.md>), [audio](<https://devfeed.tech/tags/audio.md>), [backlog](<https://devfeed.tech/tags/backlog.md>), [backpressure](<https://devfeed.tech/tags/backpressure.md>), [batch](<https://devfeed.tech/tags/batch.md>), [batch-processing](<https://devfeed.tech/tags/batch-processing.md>), [bug](<https://devfeed.tech/tags/bug.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [content](<https://devfeed.tech/tags/content.md>), [creators](<https://devfeed.tech/tags/creators.md>), [incident](<https://devfeed.tech/tags/incident.md>), [incident-report](<https://devfeed.tech/tags/incident-report.md>), [podcast](<https://devfeed.tech/tags/podcast.md>), [reliability](<https://devfeed.tech/tags/reliability.md>), [report](<https://devfeed.tech/tags/report.md>), [spotify](<https://devfeed.tech/tags/spotify.md>), [video](<https://devfeed.tech/tags/video.md>)

### AI overview

On June 24, Spotify's video transcoding system hit capacity during a publishing spike, delaying new video podcast episodes for hours. Insufficient burst headroom, a batch job processing older episodes, higher per-episode costs from a quality improvement, and a scheduling bug that reduced throughput by about 10% combined to create the backlog. Spotify stopped the batch job, fixed the scheduling bug, and added capacity; queues cleared by 01:02 UTC on June 25, and publishing was confirmed normal at 07:30. The report says capacity was later increased by approximately 67%, monitoring was improved, and a broader effort is addressing burst-aware capacity planning, priority for new creator content over background work, rate limiting and backpressure, and faster creator notifications. Early alerts preceded formal incident response by roughly four hours.

### Source excerpt

Over the past two months, podcast creators have experienced a series of reliability issues on Spotify. This... The post Content Ingestion & Podcast Video Incident Report appeared first on Spotify Engineering.