# recommendations A recommendation system uses machine-learning models to analyze user preferences and item similarities to suggest relevant content. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## pgAssistant 3.8.0 : continuous improvement loop for Postgres DevFeed: [pgAssistant 3.8.0 : continuous improvement loop for Postgres]() Original publisher: [Read original article]() Author: Pgassistant Dev Team Published: 2026-09-16T00:00:00Z Content type: release Language: en Sources: [PostgreSQL news]() Topics: [PostgreSQL](), [Monitoring](), [recommendations](), [configuration](), [real-time]() Tags: [configuration](), [measurements](), [monitoring](), [postgres](), [postgresql](), [real-time](), [recommendations](), [release]() ### AI overview pgAssistant 3.8.0 expands the PostgreSQL analysis and tuning tool into a continuous improvement platform. It adds historical workload and environment measurements, compares consecutive collections, tracks recommendations and configuration changes, and helps teams measure changes while distinguishing correlation from causation. ### Source excerpt With this release, pgAssistant is evolving beyond PostgreSQL analysis and tuning to become a continuous PostgreSQL improvement platform. The new positioning is built around a continuous improvement loop: Observe -> Diagnose -> Prioritize -> Plan -> Implement -> Collect again -> Measure pgAssistant already helped identify what should be improved and turn recommendations into a prioritized Executive Plan with clear DEV and OPS ownership. Combined with pgAssistant Collector, version 3.8.0 goes further by adding historical workload and environment measurements. The objective is to answer four essential questions: What should we improve? What did we decide to do? What did we actually change? What was the result? Workload Insights compares consecutive collections and highlights: new and no-longer-detected recommendations; changes to the PostgreSQL version and configuration; workload evolution by statement type; changes in execution time and call volume; the queries with the greatest impact on the overall workload. The ambition is to correlate the application of pgAssistant recommendations and the Executive Plan with observed performance changes. Correlation is not causation, and a recommendation that is no longer detected does not necessarily prove that it was implemented. pgAssistant keeps these distinctions explicit while bringing the relevant evidence together in one place. pgAssistant is not intended to replace real-time monitoring. Monitoring shows what is happening now; pgAssistant helps teams decide what to improve next, organize the remediation work, and measure what changed afterwards. From recommendations to action--and from action to measurable evidence. pgAssistant 3.8.0: https://github.com/beh74/pgassistant-community pgAssistant Collector: https://github.com/beh74/pgassistant-collector pgAssistant Grafana : https://github.com/beh74/pgassistant-grafana ## 10 Figma agent skills featured by Figma DevFeed: [10 Figma agent skills featured by Figma]() Original publisher: [Read original article]() Author: Miggi Cardona Published: 2026-08-13T16:00:00Z Content type: article Language: en Sources: [Figma Blog]() Topics: [Figma](), [Figma agent](), [Figma Community](), [Design system](), [recommendations]() Tags: [design-system](), [figma](), [figma-agent](), [figma-community](), [recommendations]() ### AI overview Figma explains how users can author skills for its agent directly in Figma, preview and edit them, and iterate on them through testing. The article highlights 10 skills from the Figma Community, including examples for generating radiographs and reviewing animation opportunities. ### Source excerpt Now you can author skills for our agent, directly in Figma, to put your point of view to work. Here are 10 of our favorites to get you started. ## From Scoring to Spelling: Rebuilding Ads Retrieval at Instacart DevFeed: [From Scoring to Spelling: Rebuilding Ads Retrieval at Instacart]() Original publisher: [Read original article]() Author: Karuna Ahuja Published: 2026-06-02T18:50:19Z Content type: article Language: en Sources: [Instacart]() Topics: [Transformer](), [recommendations](), [systems](), [context](), [real-time]() Tags: [bert](), [context](), [generative](), [inference](), [machine-learning](), [model](), [real-time](), [recommendation-system](), [recommendations](), [retrieval]() ### AI overview Instacart describes rebuilding its ads retrieval system by moving from a product-scoring encoder to a generative model that predicts product IDs token by token. The system uses contextual shopping sessions and a BERT-like transformer to retrieve sponsored and organic recommendations. ### Source excerpt Key Contributors: Karuna Ahuja, Marko Avdalovic, Soroush Sobhkhiz, Shrikar Archak, Xiyu Wang, Ji Chao Zhang, Hao Yan Introduction Every time a user opens Instacart, they see product recommendations: on the retailer home page, in search results, and alongside their cart. Many of these recommendations are sponsored products surfaced by a retrieval model that decides which products to show from a vast ads product catalog. A relevant ad helps users discover products they didn't know they needed; a less relevant one generates friction. Two years ago, we introduced Contextual Recommendations (CR), a BERT-based sequence model powering retrieval for both ads and organic recommendations across all major browse surfaces. In this post, we'll focus on our ads retrieval. We will detail how we rebuilt the system, by moving from an encoder that scores products to a generative model that spells them out, token by token. By doing so, we unlocked a new level of contextual matching -- ensuring brands appear exactly when users want them, while simultaneously opening up discovery of thousands of relevant products the previous system couldn't retrieve. Contextual Recommendations: A recap At its core, CR treats grocery shopping as a language modeling task, where atomic product IDs function as tokens and, the finite subset of the catalog it is trained on, acts as its 'vocabulary'. The model leverages the user's real-time session, which includes product views, item page visits, and cart additions, as a sequence of these product tokens. A BERT-like transformer is then trained on millions of authentic shopping sessions to predict the next token (i.e. singular product) in the sequence. This process allows the model to learn and capture complex purchasing patterns, such as the tendency for users who add pasta and olive oil to frequently add garlic next. This single retrieval layer replaced multiple ad-hoc systems and powers recommendation carousels across all major browse surfaces, serving both ## How Whatnot's Taxonomy Keeps Pace with a Marketplace That Won't Sit Still DevFeed: [How Whatnot's Taxonomy Keeps Pace with a Marketplace That Won't Sit Still]() Original publisher: [Read original article]() Author: Whatnot Engineering Published: 2026-03-25T14:22:19Z Content type: article Language: en Sources: [Whatnot Engineering]() Topics: [recommendations](), [bug](), [real-time]() Tags: [2025](), [ai](), [bug](), [ecommerce](), [machine-learning](), [onboarding](), [platform](), [real-time](), [recommendation-system](), [recommendations](), [taxonomy]() ### AI overview Whatnot describes how it redesigned its marketplace taxonomy to change more rapidly across countries and discovery surfaces. The article explains why onboarding, browse, search, and recommendations need different taxonomy treatments, using supply-aware onboarding and the Labubu category as examples. ### Source excerpt Marcus Gee | Discovery The hottest trends don't happen over months, but start suddenly. At Whatnot, our best sellers update their strategies in real time to stay in step with the market. Connecting buyers and sellers in this fast-changing landscape is one of the most important problems Whatnot solves, but in 2025, it was a major challenge. In this post, we're going to walk through how we overhauled the taxonomy system to evolve more rapidly, so our team could help new communities grow. There is No Single Taxonomy Whatnot supports over 200 categories across seven countries, and that number increases weekly. Our taxonomy is a platform primitive that underpins nearly every discovery experience. It shapes onboarding (what categories new users see), browse (how they navigate), search (how results are ranked), and recommendations (what their feed looks like). A bug in taxonomy doesn't break one feature; it degrades the entire discovery surface. That makes every change high-stakes. Here's what we learned the hard way: maintaining a single taxonomy doesn't work. A single taxonomy can't optimally serve onboarding, browse, and search at the same time, because these surfaces have different goals. Let's look at two examples to understand why: When new users join Whatnot, they choose categories they're interested in. That choice shapes the first feed they see. If a user selects a niche category with limited supply, their initial experience may feel sparse or low quality. For this reason, onboarding taxonomies are country-specific and supply-aware. We fold categories into broader parents, reorder them, or exclude them based on local marketplace density. Categories play a critical role in connecting enthusiasts and sellers. When Labubu surged in 2025, it was subcategorized under Toys, and new users struggled to find relevant shows. When we elevated Labubu to the first page of onboarding (a similar treatment given to other big categories like Pokémon Cards), it reduced this frictio ## Working at the intersection of data and AI with Kenza Boulisfane DevFeed: [Working at the intersection of data and AI with Kenza Boulisfane]() Original publisher: [Read original article]() Author: Thumbtack People Team Published: 2026-03-05T19:31:06Z Content type: article Language: en Sources: [Thumbtack Engineering - Medium]() Topics: [Artificial Intelligence](), [data](), [recommendations](), [real-time]() Tags: [agent](), [ai](), [analytics](), [careers](), [data](), [real-time](), [recommendations](), [software-development](), [technology]() ### AI overview In a Q&A, Thumbtack software engineer Kenza Boulisfane discusses building an AI-powered Marketing Analytics Agent. The tool is intended to help people across the company ask questions about marketing performance, receive data-backed recommendations, and explore results in real time without requiring SQL expertise. She also describes the challenges of unifying fragmented, differently defined marketing data and translating it into clear insights. ### Source excerpt Kenza Boulisfane, Software Engineer at Thumbtack, works at the intersection of data, AI, and real business impact. In this Q&A, she shares how she's building an AI-powered Marketing Analytics Agent designed to make complex marketing data accessible to everyone. She also reflects on team culture, technical challenges, and why diverse perspectives make engineering stronger. What are you currently working on? I'm working on building a Marketing Analytics Agent. It's an AI-powered marketing expert that provides companywide support, regardless of technical background. The idea is simple: marketing data shouldn't require a SQL expert to interpret it. Whether someone is in Data Science, Growth, Brand, or Product, they should be able to ask questions like, "How did paid social perform last week?" or "Why did conversions dip in this channel?" and get a clear, reliable answer. Even better, they should receive thoughtful, data-backed recommendations. Instead of waiting on reports or digging through multiple tools, teams can explore performance in real time, ask follow-up questions, and move faster with confidence. I love that my work sits at the intersection of data, AI, and real business impact. How would you describe the culture of your team? I'd describe my team culture as genuinely supportive. People are always willing to jump in, answer questions, share context, or sanity-check an idea, regardless of how busy they are. We're in it together, and that makes tackling ambitious projects feel a lot less daunting. At the same time, we don't take ourselves too seriously. Some of my favorite moments have been at offsites when we play games and you suddenly see everyone's competitive side come out. It's funny how the same people who are thoughtfully debating technical implementation during the day are fully locked in trying to win a Mafia/Werewolf round. What are some of the unique technical challenges you get to solve? One of the biggest challenges is turning fragmented marketi ## Closing the loop between user questions and documentation DevFeed: [Closing the loop between user questions and documentation]() Original publisher: [Read original article]() Author: Han Wang Published: 2026-01-07T00:00:00Z Content type: release Language: en Sources: [Mintlify Blog]() Topics: [Documentation](), [recommendations](), [pull-requests](), [dashboards](), [Artificial Intelligence]() Tags: [agents](), [ai](), [analytics](), [announcements](), [documentation](), [pull-requests](), [recommendations](), [updates]() ### AI overview Mintlify describes agent suggestions, a feature that analyzes questions users ask its documentation assistant to identify documentation gaps and recommend updates. The feature expands an existing pull-request-based system with conversation-derived signals. ### Source excerpt Learn how Mintlify now surfaces documentation updates not only from pull requests but also from real user conversations, giving teams a clearer and more complete signal of what needs to change ## Inside Glassdoor's Multi-Stage Recommendation System DevFeed: [Inside Glassdoor's Multi-Stage Recommendation System]() Original publisher: [Read original article]() Author: Prateek Shekhar Published: 2025-08-12T19:37:39Z Content type: article Language: en Sources: [Glassdoor Engineering]() Topics: [recommendation systems](), [recommendations](), [personalization](), [Algorithms](), [ordering](), [ann]() Tags: [algorithms](), [ann](), [glassdoor](), [ordering](), [personalization](), [recommendation-system](), [recommendation-systems](), [recommender-systems]() ### AI overview This article gives a high-level overview of Glassdoor's hybrid, multi-stage recommendation system. It describes how the system uses user behavior, similar-user behavior, and item features to produce personalized recommendations for the Community Feed, Email, and Push notifications, with stages including candidate generation, filtering, scoring, and ordering. ### Source excerpt Introduction: Recommendation systems are central to modern digital platforms, delivering personalized content by intelligently analyzing user behavior and preferences through advanced algorithms. Widely used in e-commerce, streaming services, social media, and online learning, they drive user satisfaction, engagement, and retention while supporting business growth. Typically, these systems operate using one or more of the following approaches: Collaborative Filtering: This approach examines user behavior by comparing how users interact with items. It identifies patterns by analyzing similar user actions. For example, if two users have watched or purchased similar items, the system will recommend those same items to other users with similar behavior. Content-Based Filtering: In this approach, the system analyzes the characteristics of the items themselves, such as text, images, or other features. It recommends items that have features similar to those a user has previously liked. For example, if a user enjoyed books with specific keywords, the system will suggest books with those exact keywords. Hybrid Approach: This approach combines collaborative and content-based filtering to enhance personalization and accuracy. It utilizes user behavior patterns in conjunction with item features, offering more personalized and accurate recommendations. At Glassdoor, our Community Feed, Email, and Push notifications are powered by a recommendation system that follows a hybrid approach. It analyzes individual users' behavior, such as posts, comments, and reactions, as well as the behavior of similar users, to deliver tailored, personalized recommendations. This post presents a high-level overview of the multi-stage architecture behind our system, highlighting candidate generation, filtering, scoring, ordering, and supporting infrastructure. Glassdoor's Recommendation System ArchitectureFigure 1: Glassdoor's Recommendation System Architecture Glassdoor's Recommendation System follo ## KotlinConf 2025 Recordings DevFeed: [KotlinConf 2025 Recordings]() Original publisher: [Read original article]() Published: 2025-07-01T00:00:00Z Content type: article Language: en Sources: [Kt. Academy]() Topics: [Kotlin](), [recommendations](), [Compiler](), [Project Valhalla](), [Coroutines](), [debugging]() Tags: [2025](), [compiler](), [coroutines](), [debugging](), [kotlinconf](), [project-valhalla](), [recommendations](), [workshop-learning-programming](), [writing]() ### AI overview An opinionated list of recommended recordings from KotlinConf 2025, including the keynote, Rich Errors in Kotlin, compiler plugins, Project Valhalla, and debugging and performance tuning in a coroutine world. ### Source excerpt My list of must-watch from KotlinConf 2025. ## Cryptographic Right Answers: Post Quantum Edition DevFeed: [Cryptographic Right Answers: Post Quantum Edition]() Original publisher: [Read original article]() Published: 2024-07-29T17:25:48Z Content type: article Language: en Sources: [Latacora]() Topics: [Post-quantum cryptography](), [Cryptography](), [recommendations]() Tags: [academic-research](), [cryptography](), [guide](), [post-quantum-cryptography]() ### AI overview This article updates cryptography recommendations for the transition to post-quantum cryptography. It explains that PQC involves more varied tradeoffs than classical cryptography and that some promising systems have failed, making recommendations harder to assess confidently. ### Source excerpt One of our favorite blog posts is our "crypto right answers" post. It's intended to be an easy-to-use guide to help engineers pick the best cryptography choices without needing to go too far down a rabbit hole. With post-quantum cryptography (PQC) recently transitioning from an academic research topic to a more practical cryptography concern we figured it's time for an update of our cryptography recommendations. One thing that makes recommending PQC challenging is that historically, we've been able to provide "better" answers for classical cryptography. Faster and bigger hashes, stronger password KDFs, easier-to-use primitives... These things all have the same fundamental "shape": you can take an existing design and drop in something else to make it better. MD5 and BLAKE3 are not comparable in strength, but you can just use BLAKE3 in place of MD5 and get something that's just far better with minimal API changes. ## How to join a new company effectively DevFeed: [How to join a new company effectively]() Original publisher: [Read original article]() Author: Tiger Abrodi Published: 2024-07-04T05:59:20Z Content type: tutorial Language: en Sources: [Saiyan Growth Letter]() Topics: [Learning](), [Users](), [meetings](), [pull-requests](), [recommendations]() Tags: [how-to](), [learning](), [meetings](), [process](), [pull-requests](), [recommendations](), [users]() ### AI overview This article offers practical advice for joining a new company, including using and understanding the product, learning about the business and its users, reviewing past issues and pull requests, meeting teammates one-on-one, and taking notes. ### Source excerpt Get up to speed and quickly start contributing in a meaningful way. ## The Culture Map DevFeed: [The Culture Map]() Original publisher: [Read original article]() Author: Tom Published: 2024-04-24T00:00:00Z Content type: article Language: en Sources: [Tomasz Gągor]() Topics: [context](), [Framework](), [recommendations]() Tags: [business](), [china](), [communication](), [context](), [culture](), [framework](), [germany](), [india](), [international](), [japan](), [leadership](), [management](), [psychology](), [recommendations](), [sociology](), [spain]() ### AI overview A discussion of Erin Meyer's The Culture Map and its framework for understanding how cultural differences affect international business. It explains high- and low-context communication and gives examples of communication, leadership, decision-making, and workplace recommendations for countries including Germany, Poland, and Spain, with additional context for Pakistan, Russia, and India. ### Source excerpt The Culture Map Breaking Through the Invisible Boundaries of Global Business Author: Erin Meyer "The Culture Map: Breaking Through the Invisible Boundaries of Global Business" by Erin Meyer is a highly insightful guide that explores how cultural differences impact international business interactions. Meyer presents a framework for understanding and navigating cultural complexities by examining various dimensions of communication, leadership, and etiquette across different countries. ## Location-Based Fashion Recommendations at Myntra DevFeed: [Location-Based Fashion Recommendations at Myntra]() Original publisher: [Read original article]() Author: Siddhartha Devapujula Published: 2024-04-22T11:25:37Z Content type: tutorial Language: en Sources: [Myntra]() Topics: [recommendation systems](), [Machine learning](), [recommendations]() Tags: [artificial-intelligence](), [ecommerce](), [fashion](), [location-based](), [machine-learning](), [personalisation](), [recommendation-system](), [recommendation-systems](), [search]() ### AI overview Myntra explains how machine-learning recommendation systems can use location as user side information to improve personalized fashion recommendations, including for cold-start users. ### Source excerpt Authored By Rohit Gupta & Siddhartha Devapujula Introduction Millions of users visit Myntra daily to upgrade their wardrobes and millions of items are listed on the platform at any given time. Users neither have the time nor the capability to scroll through this vast list of items. Even after applying category and attribute filters, usually the number of items is still in thousands. Hence it becomes critical that the top search results for any user are both relevant and personalized. Just like search, many other recommendation widgets across the platform face the same challenges. Fashion Diversity -- Every Region has its own Fashion Showing each user the best styles for them from a catalog of million plus products is where machine learning based recommendation systems come into play. From search results on google to your netflix home screen, recommendation systems are working in the background to get you the best results. It is impossible to imagine modern age internet experience without these systems. The uber goal of these models is to take the user features and the vast list of items as input ,and generate a small personalized list of items for each user. For these systems to work, we mainly use the user's historical activity on the platform. In this blog we will see how using other kinds of user details can also enhance the quality of recommendations. In the next sections, we dive into the details of recommendation systems and related techniques. We explain the motivation for a location based recommendation system and how we built one at Myntra. Later we discuss a few use cases at Myntra, results and potential future work. Basics of recommendation systems This is a very simple read about recommendation systems by Google -- Recommendations: What and Why? | Machine Learning | Google for Developers. Readers can skip if they are already aware of this. Traditional recommendation models focus on using the user's historical interactions on the platform to learn. This wor ## Implementing Multi-GPU Distributed Training for Stitch Fix's Personalized Recommendations DevFeed: [Implementing Multi-GPU Distributed Training for Stitch Fix's Personalized Recommendations]() Original publisher: [Read original article]() Published: 2023-06-08T09:00:00Z Content type: article Language: en Sources: [Stitch Fix]() Topics: [distributed-training](), [recommendations](), [GPU](), [sharding](), [Training AI Models](), [Multi-GPU](), [PyTorch]() Tags: [distributed-training](), [gpu](), [model-training](), [multi-gpu](), [parallel](), [performance](), [pytorch](), [recommendations](), [sharding]() ### AI overview This Stitch Fix engineering article explains how the company implemented multi-GPU distributed training for its Client Time Series Model (CTSM), a PyTorch-based model used in personalized recommendations. It describes sharding training data across GPUs and training mini-batches in parallel to reduce training time, along with the surrounding retraining and deployment workflow. ### Source excerpt Stitch Fix uses a cutting-edge multi-tiered recommender system stack to personalize styling recommendations at scale. This stack comprises several critical components, including feature generation, scoring, ranking, and inventory optimization techniques. Our scoring module is based on the Client Time Series Model (CTSM) which is an award winning novel sequence based model that uses temporally masked encoders. CTSM is built using PyTorch, and was initially trained on a single Graphics Processing Unit (GPU) instance. Since we first put this model into production last year, we have launched several updates to the model that improved its performance. Many of these improvements involved adding new features or increasing the time window of our training data. As a result, the model training time increased significantly, making it harder for us to iterate quickly and get feedback on new ideas we want to try for improving the model. We needed a way to reduce the model training time. This blog delves into the steps we followed to overcome this challenge and our journey to implement multi-GPU distributed model training for CTSM. By sharding the training data across multiple GPUs and training multiple mini-batches in parallel, we aimed to achieve significant reductions in training time. We present empirical results showcasing the observed reduction in training time when we scaled up resources from 1 to N GPUs, and share some future directions we are considering in our continued effort to speed up model training. Model Training Workflow The scores generated by CTSM are leveraged by multiple downstream services to get insight into what items a client is likely to purchase. The model is retrained at a regular cadence to ensure that it is using the most updated information about each client when making predictions and does not degrade in its performance. We leverage configuration driven machine learning pipelines to set up a Directed Acyclic Graph (DAG) that automatically retrains ## Client Time Series Model: a Multi-Target Recommender System based on Temporally-Masked Encoders DevFeed: [Client Time Series Model: a Multi-Target Recommender System based on Temporally-Masked Encoders]() Original publisher: [Read original article]() Published: 2022-10-14T06:00:00Z Content type: article Language: en Sources: [Stitch Fix]() Topics: [client](), [Time Series](), [recommendations](), [Embeddings](), [model architecture](), [maintenance](), [systems](), [data]() Tags: [data](), [embeddings](), [maintenance](), [model-architecture](), [recommendations](), [systems](), [time-series]() ### AI overview Stitch Fix describes its Client Time Series Model, a sequence-based recommender that estimates the probability of client-item purchases. The model uses a unified client embedding and incorporates the time dimension of client interactions to reduce duplicated models, improve maintainability, and share learning across business lines, regions, and channels. ### Source excerpt Introduction The foundation of our recommendation stack is a scoring model we call p(sale), which estimates the probability that any given client will purchase any given item. This model has gone through many iterations over the years, from a mixed effects model, to a matrix factorization model, and now to a novel sequence-based model. Internally we call this the Client Time Series Model (aka CTSM) because of its focus on the time-domain of client interactions. This post details our new model, which is a significant improvement for both the quality of our recommendations and the maintainability of our systems. Motivation Before setting out to develop our new model, it was clear that the evolution of our business necessitated a change to our modeling approach. First, the growing variety of recommendations we serve led to an explosion in the number of models we needed to maintain. Each time the business expanded, such as adding Mens, or serving the UK, or adding direct shopping with Freestyle, we responded by forking a new model to serve the new channel. This was necessary because a single domain-agnostic model could not serve the new channels as well as tailored models, but over time it has increased our maintenance burden and cost of iteration. In addition to the system complexity, we also knew we had an opportunity to make better use of important signals. With data and models separated by business line, region, and channel, we had a limited ability to leverage learning across these boundaries. With a unified model, we can more seamlessly use data from US clients to improve recommendations for UK clients, or data from Fixes to improve recommendations in Freestyle. Finally, our previous approaches modeled clients via tabular data. Although they are trained on purchase events that take place in the context of a particular point in time, they did not explicitly consider the time dimension in their understanding of the client's interactions. We believed that there was s ## Experts in-the-Loop at Stitch Fix DevFeed: [Experts in-the-Loop at Stitch Fix]() Original publisher: [Read original article]() Published: 2022-09-02T09:00:00Z Content type: article Language: en Sources: [Stitch Fix]() Topics: [personalization](), [recommendations](), [datasets](), [Training AI Models]() Tags: [customer](), [datasets](), [e-commerce](), [personalization](), [recommendations](), [training](), [training-data]() ### AI overview Stitch Fix describes how human stylists help create high-quality training data for personalized outfit recommendations in its Freestyle direct-shopping experience. The article explains why stylist judgments are needed to assess recommendation quality and client experience at scale. ### Source excerpt Imagine your job is to personalize search results on an e-commerce site for returning customers, classify the presence or absence of pedestrians in street photos, or develop an app that translates languages. In all of these cases, a basic ingredient is a dataset of annotations provided by a human. For any company seeking to personalize experience for its customers, combining human computation with algorithmic computation is essential. This is also true for Stitch Fix. At Stitch Fix, we recently launched Stitch Fix Freestyle, our direct-shopping experience, where our algorithmic recommendations are now directly shared with clients in their own personal shopping feed - a different approach from our original Fix experience, where a team of expert stylists determined what should go in the client's Fix. Central to the Freestyle experience for clients is showing individual items as part of complete outfits, where both items and outfits are personalized based on our clients unique size, fit, style, and price preferences. But, what makes a good outfit? And, how do we balance personalization to a customer with a given level of outfit quality or a particular stylistic slant? In order to bootstrap a new product like this, to offer personalized outfit recommendations at scale to our around 4m clients, we need some high-quality training data. In particular, we need data specific to what it means to have a good client experience - or what makes a good outfit for a given client at Stitch Fix. Datasets like this aren't exactly floating around: assembling them requires an intentional, large-scale effort. The best way to get high-quality data sets is to work with our in-house experts: our stylists. In this post, you'll learn more about the purpose and impact of stylist-in-the-loop projects, and the powerful impact that our stylists have in building the future of personalized shopping. Bringing in our "experts in-the-loop" Working with our expert stylists helps us improve our clients' ## Major Cyber Incidents Investigations DevFeed: [Major Cyber Incidents Investigations]() Original publisher: [Read original article]() Author: Adam Published: 2022-07-19T00:00:00Z Content type: release Language: en Sources: [Shostack & Friends Blog]() Topics: [incident](), [Cybersecurity](), [recommendations]() Tags: [cybersecurity](), [guide](), [incident](), [recommendations](), [report]() ### AI overview The article announces a report and handbook on establishing independent boards to investigate major cyber incidents. It describes the guide as a blueprint for public and private organizations and says investigations should document event sequences, contributing factors, and security-practice recommendations. ### Source excerpt I'm thrilled this how to guide for standing up new investigations is available. ## Improving Evaluation Practices in Natural Language Generation DevFeed: [Improving Evaluation Practices in Natural Language Generation]() Original publisher: [Read original article]() Author: Saad Mahamood NLG Expert; Lead Data Scientist Published: 2022-03-31T00:00:00Z Content type: article Language: en Sources: [Trivago]() Topics: [Natural language processing](), [recommendations]() Tags: [data-science](), [evaluation](), [language](), [open-source](), [recommendations](), [report](), [reproducibility](), [research]() ### AI overview This article reviews research into evaluation practices for Natural Language Generation. It discusses weaknesses in automated metrics, variation in human evaluation methods, and reproducibility concerns, including findings from trivago's HumEval 2021 work and its proposed Commonsense Evaluation Card. ### Source excerpt Throughout last year I had the opportunity to participate and collaborate on multiple research initiatives in the field of Nat... ## Learning Lessons from Aviation DevFeed: [Learning Lessons from Aviation]() Original publisher: [Read original article]() Author: Adam Published: 2021-11-15T00:00:00Z Content type: opinion Language: en Sources: [Shostack & Friends Blog]() Topics: [Security](), [recommendations](), [risk-management]() Tags: [aviation](), [investigations](), [recommendations](), [report](), [research](), [security](), [third-party]() ### AI overview The article discusses applying learning models from aviation to cybersecurity investigations. It describes a report examining what a cyber NTSB might be, recommends objective and independent investigations, and identifies research questions for further study. ### Source excerpt The definition of insanity is doing the same thing over and over and expecting different results. We can do better, and a major new report explains how. ## Real-Time Machine Learning at Headspace DevFeed: [Real-Time Machine Learning at Headspace]() Original publisher: [Read original article]() Author: Headspace Published: 2021-08-16T16:53:19Z Content type: article Language: en Sources: [Stories by Headspace on Medium]() Topics: [Machine Learning & Artificial Intelligence](), [real-time](), [Inference](), [recommendations](), [Apache Spark](), [databricks](), [Amazon Simple Queue Service (SQS)](), [Amazon Web Services](), [Amazon SageMaker AI](), [Streaming](), [Orchestration](), [Model Development]() Tags: [android](), [apache-spark](), [aws](), [data](), [databricks](), [headspace-engineering](), [inference](), [machine-learning](), [orchestration](), [real-time](), [recommendations](), [sqs](), [streaming](), [web]() ### AI overview This article describes Headspace's infrastructure for real-time machine learning inference and personalized content recommendations. It explains how the team uses modular Publishing, Receiver, Orchestration, and Serving layers with Apache Spark Structured Streaming on Databricks, AWS SQS, Lambda, and SageMaker. ### Source excerpt Author: Yu Chen, Senior Software Engineer / Co-author: Koyuki Nakamori, Senior Engineering Manager Introduction Data is often most valuable when it can be immediately leveraged to make decisions in the moment, but traditionally consumer data is ingested, transformed, persisted, and sits dormant for lengthy periods of time before being used by machine learning and analytics teams. Headspace's core products are iOS, Android, and web-based apps that focus on improving the health and happiness of its users through mindfulness, meditation, sleep, exercise, and focus content. Machine learning models are core to our user experiences by offering recommendations that engage our users with new relevant, personalized content that builds consistent habits in their lifelong journey. Finding a way to leverage user data to generate real-time insights and decisions means that consumer-facing products like the Headspace app can dramatically shorten the end-to-end user feedback loop: actions that users perform a few seconds, or minutes ago can be incorporated in generating a more relevant, personalized, and context-specific content recommendation for the user. This means our ML models could incorporate dynamic features that update throughout the course of a user's day, or even an individual session: Current session bounce rates for sleep content Semantic embeddings for recent user search terms (if a user recently searched for "preparing for big exam", the ML model can assign more weight to Focus-themed meditations) Users' biometric data (i.e., if step counts and heart rate are increasing over the last 10 minutes, we can recommend Move or Exercise content) The Headspace Machine Learning team architected a solution by decomposing the infrastructure systems into modular Publishing, Receiver, Orchestration, and Serving layers, leveraging Apache Spark Structured Streaming on Databricks, AWS SQS, Lambda, and Sagemaker to deliver real-time inference capabilities for our machine learning mod ## Encryption & Privacy Policy and Technology DevFeed: [Encryption & Privacy Policy and Technology]() Original publisher: [Read original article]() Author: Adam Published: 2019-12-11T00:00:00Z Content type: opinion Language: en Sources: [Shostack & Friends Blog]() Topics: [Encryption](), [Facial recognition](), [Cryptography](), [recommendations]() Tags: [encryption](), [facial-recognition](), [privacy](), [recommendations](), [research]() ### AI overview A commentary roundup discusses an open letter arguing that weak encryption puts internet users at risk, nominations for the Caspar Bowden Award for privacy-enhancing technology research, and proposed rules for facial recognition technology at the Port of Seattle. ### Source excerpt A few tidbits in recent news. ## Overcast 2019.6 adds per-podcast controls, longer clips, and data-driven recommendations DevFeed: [Overcast 2019.6 adds per-podcast controls, longer clips, and data-driven recommendations]() Original publisher: [Read original article]() Author: Marco Arment Published: 2019-07-21T20:14:44Z Content type: release Language: en Sources: [Marco.org]() Topics: [App](), [recommendations](), [iOS](), [data](), [macOS](), [watchOS]() Tags: [app](), [data](), [ios](), [macos](), [recommendations](), [update](), [watchos]() ### AI overview Overcast 2019.6 adds per-podcast download, streaming, and auto-delete controls, increases clip-sharing to 90 seconds, and replaces Twitter recommendations with an in-app data-driven recommendation engine. ### Source excerpt Today's Overcast update (2019.6) brings some great new features. But first, I need to set low expectations for iOS 13, watchOS 6, and macOS Catalina updates this fall. Halfway through the summer, I've made much less progress than expected, having been overwhelmed by the required OS changes, my own technical and design debts, and unusually rough betas. Rather than shipping a big iOS update, a standalone Watch app, and a Mac app on day one, these will probably come incrementally over the next year. Fortunately, I haven't stopped improving the app in the meantime. Per-podcast customization, longer clips You can now set download-vs.-stream and auto-delete behavior per podcast, one of the most frequently requested features over the last few years. (Sorry it took so long -- it required a lot of server-side work for uninteresting reasons.) I've also raised the clip-sharing length limit to 90 seconds. When I debuted this feature, I had to guess how it would be used and received, and 60 seconds seemed like a good limit. In practice, that's sometimes a bit too short to capture a complete thought. 90 seconds should be better -- most clips won't need to be that long, but it'll be nice to have the headroom when you need it. "Suggestions for You" replacing Twitter Overcast has had a Twitter-powered social-recommendations feature since version 1.0 in 2014, which made sense at the time: I didn't have any popularity data to generate recommendations from, and social networks were nicer places to be than they are today. But after five years, only 10% of active users have connected a Twitter account in Overcast, and only 0.2% of new podcast subscriptions have actually been added via Twitter recommendations. Not only was the feature not providing much value to the 10% of people who enabled it, but the other 90% of Overcast's customers haven't been getting personalized recommendations at all. In the meantime, I've accumulated five years of popularity data, so I tried writing my own recomme ## Summer Reading List DevFeed: [Summer Reading List]() Original publisher: [Read original article]() Author: Adam Published: 2018-07-31T00:00:00Z Content type: article Language: en Sources: [Shostack & Friends Blog]() Topics: [recommendations]() Tags: [blog](), [book](), [reading-list](), [recommendations]() ### AI overview A summer reading list for engineers includes the author's threat modeling book alongside works associated with Daniel Kahneman, Tony Hsieh, Nicole Forsgren, and Tom DeMarco. ### Source excerpt [no description provided} ## Personalized search with a custom Solr plugin DevFeed: [Personalized search with a custom Solr plugin]() Original publisher: [Read original article]() Author: Henrik Falch Published: 2018-04-10T06:00:00Z Content type: article Language: en Sources: [Finn.no]() Topics: [personalization](), [recommendations](), [API](), [Back end](), [Query (disambiguation)]() Tags: [api](), [backend](), [personalization](), [plugin](), [query](), [recommendation-system]() ### AI overview This article describes FINN.no's work to improve classified-ad search relevance by combining Solr scores with recommendation scores. It discusses testing a custom Solr plugin that uses an existing recommendations API and lists requirements for personalization, algorithm changes, and score balancing. ### Source excerpt Background On FINN.no people can search for classified ads, where the backend system is using Solr as the search engine. Default sorting on the vertical "torget" is by relevancy, which is based on Solr score. The Solr score for a document is again calculated from query relevance and the ad's published date. Here is an example searching for the word chair: Non-personalized search The first and third ad are bought positions, while the rest are sorted by published time and the importance of the word chair (Norwegian: stol). In the fall of 2017, we started experimenting with ways to improve the relevancy sorting. Our first try was by boosting geo distance, ads close to my position would get a higher score. But we could not see any positive changes for our product KPIs. Then we wanted to try sort by mixing scores from both Solr and our recommendation system. Our recommendation system already had an api where we could send a user id and a list of ad ids, and receive a recommendation score for each of the ads. A new solution We evaluated a few different solutions, mainly: learning to rank in Solr custom Solr plugin using the existing recommendations api Since we already had a system for recommendations, and awesome data scientists tuning the algorithms and so on, we chose to test the latter. An uncertainty was if we could get the response times needed for a search. Solr SearchComponent A Solr search component contains several phases used by the search handler. As we do not use sharded indices for the search we wanted to test, these are the important phases: prepare - Preparing the response -> parsing request parameters process - Processing the request for the current component Requirements the personalization score should influence the order of the search result we want to easily be able to change recommendation algorithm, and test a new one against the current search (with or without personalization) we also wanted to tune Solr score/recommendation score balance We theref ## Deep NLP-based Recommenders at Finn.no DevFeed: [Deep NLP-based Recommenders at Finn.no]() Original publisher: [Read original article]() Author: Simen Eide Published: 2017-09-08T13:56:49Z Content type: tutorial Language: en Sources: [Finn.no]() Topics: [Natural language processing](), [Deep learning](), [Embeddings](), [recommendations](), [Keras](), [Hackathon]() Tags: [data](), [deep-learning](), [embeddings](), [hackathon](), [keras](), [model-architecture](), [neural-networks](), [nlp](), [recommendations]() ### AI overview The article describes a FINN.no hackathon project exploring deep NLP-based recommendations for classified ads. The team used categorized ad data, word embeddings, and a convolutional neural network architecture to model similarity, but the supplied text does not include the final performance results. ### Source excerpt During a hackathon at FINN.no, we figured we wanted to learn more about deep NLP-models. FINN.no has a large database with ads of people trying to sell stuff (around 1 million active ads at any time), and they are categorized into a category tree with three or four layers. For example, full suspension bikes can be found under "Sport and outdoor activities" / "Bike sport" / "Full suspension bikes". In our daily jobs we are working on recommendations. There, we already have a content based (tf-idf) recommender build on Solr's More Like This. It seems to work well in areas where our collaborative filtering approaches does not. Would it be possible to build a deep learning NLP-model of similar performance? To achieve a measure of similarity, building a classifier of the previously mentioned categories seemed like a good choice, since we already had a lot of pre-existing data. The NLP team at Schibsted had already tokenized around six million ads as well as trained a word2vec model for us - we were ready to roll! Some preprocessing still had to be done. We ran through all ads, concatenated the title and description strings, and after a quick look at the data took the first 15 words of each ad. Model architecture proposed by the paper Our initial experiments were done with a simple "Bag of words" model included in the Keras repository, but we promptly switched over to "Convolutional Neural Networks for Sentence Classification" based architecture after hearing about it from our colleague, Tobias. By looking at the first 15 words of the ad, and using 200 dimensional embeddings for each word, our input is transformed into a 15x200 matrix. We apply three different convolutions on each document. The three convolutions looks at 2, 3 and 4 words (kernel sizes) in each convolution. It then max-pools each over the whole document, so that you end up with one value per document per convolution. For each kernel size you do 100 different filters. Finally you add a dense layer for clas [Next page]()