Just about two years ago, I started to research and read more about the applications of machine learning and deep learning techniques beyond the textbooks. Initially, it was just an enthusiasm, but soon it became a habit! I started by reading blog posts and engineering blogs published by big tech companies to have a better understanding about what is actually happening behind the scenes and how the basic ideas fit into the software engineering design components including web servers, data bases, load balancers, etc.

In this post, I gathered a growing list of such materials, where a great deal of this effort has already been done in other posts that I cited them at the end. I tried to organize these materials and posts in a logical manner that is easier to follow for somebody who already have some knowledge in ML/DL field and is interested to know more about the implementation and deployment of these concepts.

I have also prepared a similar categorization for the state of the art (SOTA) papers presented for the popular applications. Hopefully, I can continue and keep it updated.

Last update: January 2022


Table of Contents


Part-1: Systems and Concepts by Companies


Part-2: Concepts Reviews and Surveys


Part-3: Concepts In Practic

  1. Data Quality
  2. Data Engineering
  3. Data Discovery
  4. Feature Stores
  5. Classification
  6. Regression
  7. Forecasting
  8. Recommendation
  9. Search & Ranking
  10. Embeddings
  11. Natural Language Processing
  12. Sequence Modelling
  13. Computer Vision
  14. Reinforcement Learning
  15. Anomaly Detection
  16. Graph
  17. Optimization
  18. Information Extraction
  19. Weak Supervision
  20. Generation
  21. Audio
  22. Validation and A/B Testing
  23. Model Management
  24. Efficiency
  25. Ethics
  26. Infra
  27. MLOps Platforms
  28. Practices
  29. Team Structure
  30. Fails

Part-1 Machine Learning System Desing Patterns in Tech Companies


Google Machine Learning Courses πŸ”

  1. Google’s fast-paced, practical introduction to machine learning
  2. Recommendation systems
    general information regarding candidate generation (content-based and collaborative filtering), retrieval, scoring, and re-ranking
  3. Google Research Publications | Data Mining and Modeling
  4. Google Research Publications | Machine Intelligence
  5. Tensorflow embedding projector
  6. Jeff Dean On Large-Scale Deep Learning At Google

Facebook (Meta) Machine Learning Contents πŸ”

  1. Field Guide to Machine Learning video series
  2. Fighting Abuse @Scale
  3. Preventing abuse using unsupervised learning
  4. Community Standards report
  5. Scalable data classification for security and privacy
  6. Unicorn: A System for Searching the Social Graph
  7. Hive – A Petabyte Scale Data Warehouse using Hadoop
  8. Nemo: Data discovery at Facebook
  9. Embedding-based Retrieval in Facebook Search (Paper)
  10. How machine learning powers Facebook’s News Feed ranking algorithm
  11. Applied Machine Learning at Facebook: A Datacenter Infrastructure Perspective (Paper)
  12. Neural Code Search: ML-based code search using natural language queries
  13. AI advances to better detect hate speech
  14. Leveraging online social interactions for enhancing integrity at Facebook
  15. Scalable data classification for security and privacy
  16. Powered by AI: Advancing product understanding and building new shopping experiences
  17. Here’s how we’re using AI to help detect misinformation
  18. GrokNet: Unified Computer Vision Model Trunk and Embeddings For Commerce

Instagram πŸ”

  1. Powered by AI: Instagram’s Explore recommender system
  2. Lessons Learned at Instagram Stories and Feed Machine Learning
  3. Instagram Engineering

Twitter Engineering πŸ”

  1. Embeddings@Twitter
  2. Using Deep Learning at Scale in Twitter’s Timelines
  3. Improving engagement on digital ads with delayed feedback

Uber πŸ”

  1. Forecasting at Uber: An Introduction
  2. Applying Customer Feedback: How NLP & Deep Learning Improve Uber’s Maps
  3. Food Discovery with Uber Eats: Building a Query Understanding Engine
  4. Food Discovery with Uber Eats: Recommending for the Marketplace
  5. Food Discovery with Uber Eats: Using Graph Learning to Power Recommendations

Netflix πŸ”

  1. [Video] A Multi-Armed Bandit Framework for Recommendations at Netflix | Netflix

LinkedIn πŸ”

  1. An Introduction to AI at LinkedIn
  2. Communities AI: Building Communities Around Interests on LinkedIn
  3. Fairness, Privacy, and Transparency by Design in AI/ML Systems

Airbnb πŸ”

  1. Using Machine Learning to Predict Value of Homes On Airbnb
  2. Listing Embeddings in Search Ranking
  3. Learning Market Dynamics for Optimal Pricing
  4. Categorizing Listing Photos at Airbnb
  5. Applying Deep Learning To Airbnb Search
  6. Discovering and Classifying In-app Message Intent at Airbnb
  7. Machine Learning-Powered Search Ranking of Airbnb Experiences

Spotify πŸ”

  1. Personalized music recommendations
  2. For Your Ears Only: Personalizing Spotify Home with Machine Learning

Part-2: Machine Learning Desing Concept | Reviews and Surveys


Recommendation πŸ”

Deep Learning πŸ”

Natural Language Processing πŸ”

Computer Vision πŸ”

Vision and Language πŸ”

Reinforcement Learning πŸ”

Graph πŸ”

Embeddings πŸ”

Meta-learning and Few-shot Learning πŸ”

Others πŸ”


Part-3: Machine Learning Desing Concepts | By Topics


Data Quality πŸ”

  1. Reliable and Scalable Data Ingestion at Airbnb βž– Airbnb 2016
  2. Monitoring Data Quality at Scale with Statistical Modeling βž– Uber 2017
  3. Data Management Challenges in Production Machine Learning (Paper) βž– Google 2017
  4. Automating Large-Scale Data Quality Verification (Paper)Amazon 2018
  5. Meet Hodor β€” Gojek’s Upstream Data Quality Tool βž– Gojek 2019
  6. An Approach to Data Quality for Netflix Personalization Systems βž– Netflix 2020
  7. Improving Accuracy By Certainty Estimation of Human Decisions, Labels, and Raters (Paper) βž– Facebook 2020

Data Engineering πŸ”

  1. Zipline: Airbnb’s Machine Learning Data Management Platform βž– Airbnb 2018
  2. Sputnik: Airbnb’s Apache Spark Framework for Data Engineering βž– Airbnb 2020
  3. Unbundling Data Science Workflows with Metaflow and AWS Step Functions βž– Netflix 2020
  4. How DoorDash is Scaling its Data Platform to Delight Customers and Meet Growing Demand βž– DoorDash 2020
  5. Revolutionizing Money Movements at Scale with Strong Data Consistency βž– Uber 2020
  6. Zipline - A Declarative Feature Engineering Framework βž– Airbnb 2020
  7. Automating Data Protection at Scale, Part 1 (Part 2) βž– Airbnb 2021
  8. Real-time Data Infrastructure at Uber βž– Uber 2021
  9. Introducing Fabricator: A Declarative Feature Engineering Framework βž– Doordash 2022
  10. Functions & DAGs: introducing Hamilton, a microframework for dataframe generation βž– Stitch Fix 2021

Data Discovery πŸ”

  1. Apache Atlas: Data Goverance and Metadata Framework for Hadoop (Code) βž– Apache
  2. Collect, Aggregate, and Visualize a Data Ecosystem’s Metadata (Code) βž– WeWork
  3. Discovery and Consumption of Analytics Data at Twitter βž– Twitter 2016
  4. Democratizing Data at Airbnb βž– Airbnb 2017
  5. Databook: Turning Big Data into Knowledge with Metadata at Uber βž– Uber 2018
  6. Metacat: Making Big Data Discoverable and Meaningful at Netflix (Code) βž– Netflix 2018
  7. Amundsen β€” Lyft’s Data Discovery & Metadata Engine βž– Lyft 2019
  8. Open Sourcing Amundsen: A Data Discovery And Metadata Platform (Code) βž– Lyft 2019
  9. DataHub: A Generalized Metadata Search & Discovery Tool (Code) βž– LinkedIn 2019
  10. Amundsen: One Year Later βž– Lyft 2020
  11. Using Amundsen to Support User Privacy via Metadata Collection at Square βž– Square 2020
  12. Turning Metadata Into Insights with Databook βž– Uber 2020
  13. DataHub: Popular Metadata Architectures Explained βž– LinkedIn 2020
  14. How We Improved Data Discovery for Data Scientists at Spotify βž– Spotify 2020
  15. How We’re Solving Data Discovery Challenges at Shopify βž– Shopify 2020
  16. Nemo: Data discovery at Facebook βž– Facebook 2020
  17. Exploring Data @ Netflix (Code) βž– Netflix 2021

Feature Stores πŸ”

  1. Distributed Time Travel for Feature Generation βž– Netflix 2016
  2. Building the Activity Graph, Part 2 (Feature Storage Section) βž– LinkedIn 2017
  3. Fact Store at Scale for Netflix Recommendations βž– Netflix 2018
  4. Zipline: Airbnb’s Machine Learning Data Management Platform βž– Airbnb 2018
  5. Introducing Feast: An Open Source Feature Store for Machine Learning (Code) βž– Gojek 2019
  6. Michelangelo Palette: A Feature Engineering Platform at Uber βž– Uber 2019
  7. The Architecture That Powers Twitter’s Feature Store βž– Twitter 2019
  8. Accelerating Machine Learning with the Feature Store Service βž– CondΓ© Nast 2019
  9. Feast: Bridging ML Models and Data βž– Gojek 2020
  10. Building a Scalable ML Feature Store with Redis, Binary Serialization, and Compression βž– DoorDash 2020
  11. Rapid Experimentation Through Standardization: Typed AI features for LinkedIn’s Feed βž– LinkedIn 2020
  12. Building a Feature Store βž– Monzo Bank 2020
  13. Butterfree: A Spark-based Framework for Feature Store Building (Code) βž– QuintoAndar 2020
  14. Building Riviera: A Declarative Real-Time Feature Engineering Framework βž– DoorDash 2021
  15. Optimal Feature Discovery: Better, Leaner Machine Learning Models Through Information Theory βž– Uber 2021
  16. ML Feature Serving Infrastructure at Lyft βž– Lyft 2021

Classification πŸ”

  1. Prediction of Advertiser Churn for Google AdWords (Paper) βž– Google 2010
  2. High-Precision Phrase-Based Document Classification on a Modern Scale (Paper) βž– LinkedIn 2011
  3. Chimera: Large-scale Classification using Machine Learning, Rules, and Crowdsourcing (Paper) βž– Walmart 2014
  4. Large-scale Item Categorization in e-Commerce Using Multiple Recurrent Neural Networks (Paper) βž– NAVER 2016
  5. Learning to Diagnose with LSTM Recurrent Neural Networks (Paper) βž– Google 2017
  6. Discovering and Classifying In-app Message Intent at Airbnb βž– Airbnb 2019
  7. Teaching Machines to Triage Firefox Bugs βž– Mozilla 2019
  8. Categorizing Products at Scale βž– Shopify 2020
  9. How We Built the Good First Issues Feature βž– GitHub 2020
  10. Testing Firefox More Efficiently with Machine Learning βž– Mozilla 2020
  11. Using ML to Subtype Patients Receiving Digital Mental Health Interventions (Paper) βž– Microsoft 2020
  12. Scalable Data Classification for Security and Privacy (Paper) βž– Facebook 2020
  13. Uncovering Online Delivery Menu Best Practices with Machine Learning βž– DoorDash 2020
  14. Using a Human-in-the-Loop to Overcome the Cold Start Problem in Menu Item Tagging βž– DoorDash 2020
  15. Deep Learning: Product Categorization and Shelving βž– Walmart 2021
  16. Large-scale Item Categorization for e-Commerce (Paper) βž– DianPing, eBay 2021

Regression πŸ”

  1. Using Machine Learning to Predict Value of Homes On Airbnb βž– Airbnb 2017
  2. Using Machine Learning to Predict the Value of Ad Requests βž– Twitter 2020
  3. Open-Sourcing Riskquant, a Library for Quantifying Risk (Code) βž– Netflix 2020
  4. Solving for Unobserved Data in a Regression Model Using a Simple Data Adjustment βž– DoorDash 2020

Forecasting πŸ”

  1. Engineering Extreme Event Forecasting at Uber with RNN βž– Uber 2017
  2. Forecasting at Uber: An Introduction βž– Uber 2018
  3. Transforming Financial Forecasting with Data Science and Machine Learning at Uber βž– Uber 2018
  4. Under the Hood of Gojek’s Automated Forecasting Tool βž– Gojek 2019
  5. BusTr: Predicting Bus Travel Times from Real-Time Traffic (Paper, Video) βž– Google 2020
  6. Retraining Machine Learning Models in the Wake of COVID-19 βž– DoorDash 2020
  7. Automatic Forecasting using Prophet, Databricks, Delta Lake and MLflow (Paper, Code) βž– Atlassian 2020
  8. Introducing Orbit, An Open Source Package for Time Series Inference and Forecasting (Paper, Video, Code) βž– Uber 2021
  9. Managing Supply and Demand Balance Through Machine Learning βž– DoorDash 2021
  10. Greykite: A flexible, intuitive, and fast forecasting library βž– LinkedIn 2021

Recommendation πŸ”

  1. Amazon.com Recommendations: Item-to-Item Collaborative Filtering (Paper) βž– Amazon 2003
  2. Netflix Recommendations: Beyond the 5 stars (Part 1 (Part 2) βž– Netflix 2012
  3. How Music Recommendation Works β€” And Doesn’t Work βž– Spotify 2012
  4. Learning to Rank Recommendations with the k -Order Statistic Loss (Paper) βž– Google 2013
  5. Recommending Music on Spotify with Deep Learning βž– Spotify 2014
  6. Learning a Personalized Homepage βž– Netflix 2015
  7. Session-based Recommendations with Recurrent Neural Networks (Paper) βž– Telefonica 2016
  8. Deep Neural Networks for YouTube Recommendations βž– YouTube 2016
  9. E-commerce in Your Inbox: Product Recommendations at Scale (Paper) βž– Yahoo 2016
  10. To Be Continued: Helping you find shows to continue watching on Netflix βž– Netflix 2016
  11. Personalized Recommendations in LinkedIn Learning βž– LinkedIn 2016
  12. Personalized Channel Recommendations in Slack βž– Slack 2016
  13. Recommending Complementary Products in E-Commerce Push Notifications (Paper) βž– Alibaba 2017
  14. Artwork Personalization at Netflix βž– Netflix 2017
  15. A Meta-Learning Perspective on Cold-Start Recommendations for Items (Paper) βž– Twitter 2017
  16. Pixie: A System for Recommending 3+ Billion Items to 200+ Million Users in Real-Time (Paper) βž– Pinterest 2017
  17. How 20th Century Fox uses ML to predict a movie audience (Paper) βž– 20th Century Fox 2018
  18. Calibrated Recommendations (Paper) βž– Netflix 2018
  19. Food Discovery with Uber Eats: Recommending for the Marketplace βž– Uber 2018
  20. Explore, Exploit, and Explain: Personalizing Explainable Recommendations with Bandits (Paper) βž– Spotify 2018
  21. Behavior Sequence Transformer for E-commerce Recommendation in Alibaba (Paper) βž– Alibaba 2019
  22. SDM: Sequential Deep Matching Model for Online Large-scale Recommender System (Paper) βž– Alibaba 2019
  23. Multi-Interest Network with Dynamic Routing for Recommendation at Tmall (Paper) βž– Alibaba 2019
  24. Personalized Recommendations for Experiences Using Deep Learning βž– TripAdvisor 2019
  25. Powered by AI: Instagram’s Explore recommender system βž– Facebook 2019
  26. Marginal Posterior Sampling for Slate Bandits (Paper) βž– Netflix 2019
  27. Food Discovery with Uber Eats: Using Graph Learning to Power Recommendations βž– Uber 2019
  28. Music recommendation at Spotify βž– Spotify 2019
  29. Using Machine Learning to Predict what File you Need Next (Part 1) βž– Dropbox 2019
  30. Using Machine Learning to Predict what File you Need Next (Part 2) βž– Dropbox 2019
  31. Learning to be Relevant: Evolution of a Course Recommendation System βž– LinkedIn 2019
  32. Temporal-Contextual Recommendation in Real-Time (Paper) βž– Amazon 2020
  33. P-Companion: A Framework for Diversified Complementary Product Recommendation (Paper) βž– Amazon 2020
  34. Deep Interest with Hierarchical Attention Network for Click-Through Rate Prediction (Paper) βž– Alibaba 2020
  35. TPG-DNN: A Method for User Intent Prediction with Multi-task Learning (Paper) βž– Alibaba 2020
  36. PURS: Personalized Unexpected Recommender System for Improving User Satisfaction (Paper) βž– Alibaba 2020
  37. Controllable Multi-Interest Framework for Recommendation (Paper) βž– Alibaba 2020
  38. MiNet: Mixed Interest Network for Cross-Domain Click-Through Rate Prediction (Paper) βž– Alibaba 2020
  39. ATBRG: Adaptive Target-Behavior Relational Graph Network for Effective Recommendation (Paper) βž– Alibaba 2020
  40. For Your Ears Only: Personalizing Spotify Home with Machine Learning βž– Spotify 2020
  41. Reach for the Top: How Spotify Built Shortcuts in Just Six Months βž– Spotify 2020
  42. Contextual and Sequential User Embeddings for Large-Scale Music Recommendation (Paper) βž– Spotify 2020
  43. The Evolution of Kit: Automating Marketing Using Machine Learning βž– Shopify 2020
  44. A Closer Look at the AI Behind Course Recommendations on LinkedIn Learning (Part 1) βž– LinkedIn 2020
  45. A Closer Look at the AI Behind Course Recommendations on LinkedIn Learning (Part 2) βž– LinkedIn 2020
  46. Building a Heterogeneous Social Network Recommendation System βž– LinkedIn 2020
  47. How TikTok recommends videos #ForYou βž– ByteDance 2020
  48. Zero-Shot Heterogeneous Transfer Learning from RecSys to Cold-Start Search Retrieval (Paper) βž– Google 2020
  49. Improved Deep & Cross Network for Feature Cross Learning in Web-scale LTR Systems (Paper) βž– Google 2020
  50. Mixed Negative Sampling for Learning Two-tower Neural Networks in Recommendations (Paper) βž– Google 2020
  51. Future Data Helps Training: Modeling Future Contexts for Session-based Recommendation (Paper) βž– Tencent 2020
  52. A Case Study of Session-based Recommendations in the Home-improvement Domain (Paper) βž– Home Depot 2020
  53. Balancing Relevance and Discovery to Inspire Customers in the IKEA App (Paper) βž– Ikea 2020
  54. How we use AutoML, Multi-task learning and Multi-tower models for Pinterest Ads βž– Pinterest 2020
  55. Multi-task Learning for Related Products Recommendations at Pinterest βž– Pinterest 2020
  56. Improving the Quality of Recommended Pins with Lightweight Ranking βž– Pinterest 2020
  57. Personalized Cuisine Filter Based on Customer Preference and Local Popularity βž– DoorDash 2020
  58. How We Built a Matchmaking Algorithm to Cross-Sell Products βž– Gojek 2020
  59. Lessons Learned Addressing Dataset Bias in Model-Based Candidate Generation (Paper) βž– Twitter 2021
  60. Self-supervised Learning for Large-scale Item Recommendations (Paper) βž– Google 2021
  61. Deep Retrieval: End-to-End Learnable Structure Model for Large-Scale Recommendations (Paper) βž– ByteDance 2021
  62. Using AI to Help Health Experts Address the COVID-19 Pandemic βž– Facebook 2021
  63. Advertiser Recommendation Systems at Pinterest βž– Pinterest 2021
  64. On YouTube’s Recommendation System βž– YouTube 2021

Search & Ranking πŸ”

  1. Amazon Search: The Joy of Ranking Products (Paper, Video, Code) βž– Amazon 2016
  2. How Lazada Ranks Products to Improve Customer Experience and Conversion βž– Lazada 2016
  3. Ranking Relevance in Yahoo Search (Paper) βž– Yahoo 2016
  4. Learning to Rank Personalized Search Results in Professional Networks (Paper) βž– LinkedIn 2016
  5. Using Deep Learning at Scale in Twitter’s Timelines βž– Twitter 2017
  6. An Ensemble-based Approach to Click-Through Rate Prediction for Promoted Listings at Etsy (Paper) βž– Etsy 2017
  7. Powering Search & Recommendations at DoorDash βž– DoorDash 2017
  8. Applying Deep Learning To Airbnb Search (Paper) βž– Airbnb 2018
  9. In-session Personalization for Talent Search (Paper) βž– LinkedIn 2018
  10. Talent Search and Recommendation Systems at LinkedIn (Paper) βž– LinkedIn 2018
  11. Food Discovery with Uber Eats: Building a Query Understanding Engine βž– Uber 2018
  12. Globally Optimized Mutual Influence Aware Ranking in E-Commerce Search (Paper) βž– Alibaba 2018
  13. Reinforcement Learning to Rank in E-Commerce Search Engine (Paper) βž– Alibaba 2018
  14. Semantic Product Search (Paper) βž– Amazon 2019
  15. Machine Learning-Powered Search Ranking of Airbnb Experiences βž– Airbnb 2019
  16. Entity Personalized Talent Search Models with Tree Interaction Features (Paper) βž– LinkedIn 2019
  17. The AI Behind LinkedIn Recruiter Search and recommendation systems βž– LinkedIn 2019
  18. Learning Hiring Preferences: The AI Behind LinkedIn Jobs βž– LinkedIn 2019
  19. The Secret Sauce Behind Search Personalisation βž– Gojek 2019
  20. Neural Code Search: ML-based Code Search Using Natural Language Queries βž– Facebook 2019
  21. Aggregating Search Results from Heterogeneous Sources via Reinforcement Learning (Paper) βž– Alibaba 2019
  22. Cross-domain Attention Network with Wasserstein Regularizers for E-commerce Search βž– Alibaba 2019
  23. Understanding Searches Better Than Ever Before (Paper) βž– Google 2019
  24. How We Used Semantic Search to Make Our Search 10x Smarter βž– Tokopedia 2019
  25. Query2vec: Search query expansion with query embeddings βž– GrubHub 2019
  26. MOBIUS: Towards the Next Generation of Query-Ad Matching in Baidu’s Sponsored Search βž– Baidu 2019
  27. Why Do People Buy Seemingly Irrelevant Items in Voice Product Search? (Paper) βž– Amazon 2020
  28. Managing Diversity in Airbnb Search (Paper) βž– Airbnb 2020
  29. Improving Deep Learning for Airbnb Search (Paper) βž– Airbnb 2020
  30. Quality Matches Via Personalized AI for Hirer and Seeker Preferences βž– LinkedIn 2020
  31. Understanding Dwell Time to Improve LinkedIn Feed Ranking βž– LinkedIn 2020
  32. Ads Allocation in Feed via Constrained Optimization (Paper, Video) βž– LinkedIn 2020
  33. Understanding Dwell Time to Improve LinkedIn Feed Ranking βž– LinkedIn 2020
  34. AI at Scale in Bing βž– Microsoft 2020
  35. Query Understanding Engine in Traveloka Universal Search βž– Traveloka 2020
  36. Bayesian Product Ranking at Wayfair βž– Wayfair 2020
  37. COLD: Towards the Next Generation of Pre-Ranking System (Paper) βž– Alibaba 2020
  38. Shop The Look: Building a Large Scale Visual Shopping System at Pinterest (Paper, Video) βž– Pinterest 2020
  39. Driving Shopping Upsells from Pinterest Search βž– Pinterest 2020
  40. GDMix: A Deep Ranking Personalization Framework (Code) βž– LinkedIn 2020
  41. Bringing Personalized Search to Etsy βž– Etsy 2020
  42. Building a Better Search Engine for Semantic Scholar βž– Allen Institute for AI 2020
  43. Query Understanding for Natural Language Enterprise Search (Paper) βž– Salesforce 2020
  44. Things Not Strings: Understanding Search Intent with Better Recall βž– DoorDash 2020
  45. Query Understanding for Surfacing Under-served Music Content (Paper) βž– Spotify 2020
  46. Embedding-based Retrieval in Facebook Search (Paper) βž– Facebook 2020
  47. Towards Personalized and Semantic Retrieval for E-commerce Search via Embedding Learning (Paper) βž– JD 2020
  48. QUEEN: Neural query rewriting in e-commerce (Paper) βž– Amazon 2021
  49. Using Learning-to-rank to Precisely Locate Where to Deliver Packages (Paper) βž– Amazon 2021
  50. Seasonal relevance in e-commerce search (Paper) βž– Amazon 2021
  51. Graph Intention Network for Click-through Rate Prediction in Sponsored Search (Paper) βž– Alibaba 2021
  52. How We Built A Context-Specific Bidding System for Etsy Ads βž– Etsy 2021
  53. Pre-trained Language Model based Ranking in Baidu Search (Paper) βž– Baidu 2021
  54. Stitching together spaces for query-based recommendations βž– Stitch Fix 2021
  55. Deep Natural Language Processing for LinkedIn Search Systems (Paper) βž– LinkedIn 2021
  56. Siamese BERT-based Model for Web Search Relevance Ranking Evaluated on a New Czech Dataset (Paper, Code) βž– Seznam 2021

Embeddings πŸ”

  1. Vector Representation Of Items, Customer And Cart To Build A Recommendation System (Paper) βž– Sears 2017
  2. Billion-scale Commodity Embedding for E-commerce Recommendation in Alibaba (Paper) βž– Alibaba 2018
  3. Embeddings@Twitter βž– Twitter 2018
  4. Listing Embeddings in Search Ranking (Paper) βž– Airbnb 2018
  5. Understanding Latent Style βž– Stitch Fix 2018
  6. Towards Deep and Representation Learning for Talent Search at LinkedIn (Paper) βž– LinkedIn 2018
  7. Personalized Store Feed with Vector Embeddings βž– DoorDash 2018
  8. Should we Embed? A Study on Performance of Embeddings for Real-Time Recommendations(Paper) βž– Moshbit 2019
  9. Machine Learning for a Better Developer Experience βž– Netflix 2020
  10. Announcing ScaNN: Efficient Vector Similarity Search (Paper, Code) βž– Google 2020
  11. Embedding-based Retrieval at Scribd βž– Scribd 2021

Natural Language Processing πŸ”

  1. Abusive Language Detection in Online User Content (Paper) βž– Yahoo 2016
  2. Smart Reply: Automated Response Suggestion for Email (Paper) βž– Google 2016
  3. Building Smart Replies for Member Messages βž– LinkedIn 2017
  4. How Natural Language Processing Helps LinkedIn Members Get Support Easily βž– LinkedIn 2019
  5. Gmail Smart Compose: Real-Time Assisted Writing (Paper) βž– Google 2019
  6. Goal-Oriented End-to-End Conversational Models with Profile Features in a Real-World Setting (Paper) βž– Amazon 2019
  7. Give Me Jeans not Shoes: How BERT Helps Us Deliver What Clients Want βž– Stitch Fix 2019
  8. DeText: A deep NLP Framework for Intelligent Text Understanding (Code) βž– LinkedIn 2020
  9. SmartReply for YouTube Creators βž– Google 2020
  10. Using Neural Networks to Find Answers in Tables (Paper) βž– Google 2020
  11. A Scalable Approach to Reducing Gender Bias in Google Translate βž– Google 2020
  12. Assistive AI Makes Replying Easier βž– Microsoft 2020
  13. AI Advances to Better Detect Hate Speech βž– Facebook 2020
  14. A State-of-the-Art Open Source Chatbot (Paper) βž– Facebook 2020
  15. A Highly Efficient, Real-Time Text-to-Speech System Deployed on CPUs βž– Facebook 2020
  16. Deep Learning to Translate Between Programming Languages (Paper, Code) βž– Facebook 2020
  17. Deploying Lifelong Open-Domain Dialogue Learning (Paper) βž– Facebook 2020
  18. Introducing Dynabench: Rethinking the way we benchmark AI βž– Facebook 2020
  19. How Gojek Uses NLP to Name Pickup Locations at Scale βž– Gojek 2020
  20. The State-of-the-art Open-Domain Chatbot in Chinese and English (Paper) βž– Baidu 2020
  21. PEGASUS: A State-of-the-Art Model for Abstractive Text Summarization (Paper, Code) βž– Google 2020
  22. Photon: A Robust Cross-Domain Text-to-SQL System (Paper) (Demo) βž– Salesforce 2020
  23. GeDi: A Powerful New Method for Controlling Language Models (Paper, Code) βž– Salesforce 2020
  24. Applying Topic Modeling to Improve Call Center Operations βž– RICOH 2020
  25. WIDeText: A Multimodal Deep Learning Framework βž– Airbnb 2020
  26. Dynaboard: Moving Beyond Accuracy to Holistic Model Evaluation in NLP (Code) βž– Facebook 2021
  27. How we reduced our text similarity runtime by 99.96% βž– Microsoft 2021
  28. Textless NLP: Generating expressive speech from raw audio (Part 1) (Part 2) (Part 3) (Code and Pretrained Models) βž– Facebook 2021

Sequence Modelling πŸ”

  1. Doctor AI: Predicting Clinical Events via Recurrent Neural Networks (Paper) βž– Sutter Health 2015
  2. Deep Learning for Understanding Consumer Histories (Paper) βž– Zalando 2016
  3. Using Recurrent Neural Network Models for Early Detection of Heart Failure Onset (Paper) βž– Sutter Health 2016
  4. Continual Prediction of Notification Attendance with Classical and Deep Networks (Paper) βž– Telefonica 2017
  5. Deep Learning for Electronic Health Records (Paper) βž– Google 2018
  6. Practice on Long Sequential User Behavior Modeling for Click-Through Rate Prediction (Paper)Alibaba 2019
  7. Search-based User Interest Modeling with Sequential Behavior Data for CTR Prediction (Paper) βž– Alibaba 2020
  8. How Duolingo uses AI in every part of its app βž– Duolingo 2020
  9. Leveraging Online Social Interactions For Enhancing Integrity at Facebook (Paper, Video) βž– Facebook 2020
  10. Using deep learning to detect abusive sequences of member activity (Video) βž– LinkedIn 2021

Computer Vision πŸ”

  1. Creating a Modern OCR Pipeline Using Computer Vision and Deep Learning βž– Dropbox 2017
  2. Categorizing Listing Photos at Airbnb βž– Airbnb 2018
  3. Amenity Detection and Beyond β€” New Frontiers of Computer Vision at Airbnb βž– Airbnb 2019
  4. How we Improved Computer Vision Metrics by More Than 5% Only by Cleaning Labelling Errors βž– Deepomatic
  5. Making machines recognize and transcribe conversations in meetings using audio and video βž– Microsoft 2019
  6. Powered by AI: Advancing product understanding and building new shopping experiences βž– Facebook 2020
  7. A Neural Weather Model for Eight-Hour Precipitation Forecasting (Paper) βž– Google 2020
  8. Machine Learning-based Damage Assessment for Disaster Relief (Paper) βž– Google 2020
  9. RepNet: Counting Repetitions in Videos (Paper) βž– Google 2020
  10. Converting Text to Images for Product Discovery (Paper) βž– Amazon 2020
  11. How Disney Uses PyTorch for Animated Character Recognition βž– Disney 2020
  12. Image Captioning as an Assistive Technology (Video) βž– IBM 2020
  13. AI for AG: Production machine learning for agriculture βž– Blue River 2020
  14. AI for Full-Self Driving at Tesla βž– Tesla 2020
  15. On-device Supermarket Product Recognition βž– Google 2020
  16. Using Machine Learning to Detect Deficient Coverage in Colonoscopy Screenings (Paper) βž– Google 2020
  17. Shop The Look: Building a Large Scale Visual Shopping System at Pinterest (Paper, Video) βž– Pinterest 2020
  18. Developing Real-Time, Automatic Sign Language Detection for Video Conferencing (Paper) βž– Google 2020
  19. Vision-based Price Suggestion for Online Second-hand Items (Paper) βž– Alibaba 2020
  20. New AI Research to Help Predict COVID-19 Resource Needs From X-rays (Paper, Model) βž– Facebook 2021
  21. An Efficient Training Approach for Very Large Scale Face Recognition (Paper) βž– Alibaba 2021
  22. Identifying Document Types at Scribd βž– Scribd 2021
  23. Semi-Supervised Visual Representation Learning for Fashion Compatibility (Paper) βž– Walmart 2021

Reinforcement Learning πŸ”

  1. Deep Reinforcement Learning for Sponsored Search Real-time Bidding (Paper) βž– Alibaba 2018
  2. Budget Constrained Bidding by Model-free Reinforcement Learning in Display Advertising (Paper) βž– Alibaba 2018
  3. Reinforcement Learning for On-Demand Logistics βž– DoorDash 2018
  4. Reinforcement Learning to Rank in E-Commerce Search Engine (Paper) βž– Alibaba 2018
  5. Dynamic Pricing on E-commerce Platform with Deep Reinforcement Learning (Paper) βž– Alibaba 2019
  6. Productionizing Deep Reinforcement Learning with Spark and MLflow βž– Zynga 2020
  7. Deep Reinforcement Learning in Production Part1 Part 2 βž– Zynga 2020
  8. Building AI Trading Systems βž– Denny Britz 2020

Anomaly Detection πŸ”

  1. Detecting Performance Anomalies in External Firmware Deployments βž– Netflix 2019
  2. Detecting and Preventing Abuse on LinkedIn using Isolation Forests (Code) βž– LinkedIn 2019
  3. Deep Anomaly Detection with Spark and Tensorflow (Hopsworks Video) βž– Swedbank, Hopsworks 2019
  4. Preventing Abuse Using Unsupervised Learning βž– LinkedIn 2020
  5. The Technology Behind Fighting Harassment on LinkedIn βž– LinkedIn 2020
  6. Uncovering Insurance Fraud Conspiracy with Network Learning (Paper) βž– Ant Financial 2020
  7. How Does Spam Protection Work on Stack Exchange? βž– Stack Exchange 2020
  8. Auto Content Moderation in C2C e-Commerce βž– Mercari 2020
  9. Blocking Slack Invite Spam With Machine Learning βž– Slack 2020
  10. Cloudflare Bot Management: Machine Learning and More βž– Cloudflare 2020
  11. Anomalies in Oil Temperature Variations in a Tunnel Boring Machine βž– SENER 2020
  12. Using Anomaly Detection to Monitor Low-Risk Bank Customers βž– Rabobank 2020
  13. Fighting fraud with Triplet Loss βž– OLX Group 2020
  14. Facebook is Now Using AI to Sort Content for Quicker Moderation (Alternative) βž– Facebook 2020
  15. How AI is getting better at detecting hate speech Part 1, Part 2, Part 3, Part 4 βž– Facebook 2020

Graph πŸ”

  1. Building The LinkedIn Knowledge Graph βž– LinkedIn 2016
  2. Scaling Knowledge Access and Retrieval at Airbnb βž– Airbnb 2018
  3. Graph Convolutional Neural Networks for Web-Scale Recommender Systems (Paper)Pinterest 2018
  4. Food Discovery with Uber Eats: Using Graph Learning to Power Recommendations βž– Uber 2019
  5. AliGraph: A Comprehensive Graph Neural Network Platform (Paper) βž– Alibaba 2019
  6. Contextualizing Airbnb by Building Knowledge Graph βž– Airbnb 2019
  7. Retail Graph β€” Walmart’s Product Knowledge Graph βž– Walmart 2020
  8. Traffic Prediction with Advanced Graph Neural Networks βž– DeepMind 2020
  9. SimClusters: Community-Based Representations for Recommendations (Paper, Video) βž– Twitter 2020
  10. Metapaths guided Neighbors aggregated Network for Heterogeneous Graph Reasoning (Paper) βž– Alibaba 2021
  11. Graph Intention Network for Click-through Rate Prediction in Sponsored Search (Paper) βž– Alibaba 2021
  12. JEL: Applying End-to-End Neural Entity Linking in JPMorgan Chase (Paper) βž– JPMorgan Chase 2021

Optimization πŸ”

  1. Matchmaking in Lyft Line (Part 1) (Part 2) (Part 3) βž– Lyft 2016
  2. The Data and Science behind GrabShare Carpooling (Part 1) βž– Grab 2017
  3. How Trip Inferences and Machine Learning Optimize Delivery Times on Uber Eats βž– Uber 2018
  4. Next-Generation Optimization for Dasher Dispatch at DoorDash βž– DoorDash 2020
  5. Optimization of Passengers Waiting Time in Elevators Using Machine Learning βž– Thyssen Krupp AG 2020
  6. Think Out of The Package: Recommending Package Types for E-commerce Shipments (Paper) βž– Amazon 2020
  7. Optimizing DoorDash’s Marketing Spend with Machine Learning βž– DoorDash 2020

Information Extraction πŸ”

  1. Unsupervised Extraction of Attributes and Their Values from Product Description (Paper) βž– Rakuten 2013
  2. Using Machine Learning to Index Text from Billions of Images βž– Dropbox 2018
  3. Extracting Structured Data from Templatic Documents (Paper) βž– Google 2020
  4. AutoKnow: self-driving knowledge collection for products of thousands of types (Paper, Video) βž– Amazon 2020
  5. One-shot Text Labeling using Attention and Belief Propagation for Information Extraction (Paper) βž– Alibaba 2020
  6. Information Extraction from Receipts with Graph Convolutional Networks βž– Nanonets 2021

Weak Supervision πŸ”

  1. Snorkel DryBell: A Case Study in Deploying Weak Supervision at Industrial Scale (Paper) βž– Google 2019
  2. Osprey: Weak Supervision of Imbalanced Extraction Problems without Code (Paper) βž– Intel 2019
  3. Overton: A Data System for Monitoring and Improving Machine-Learned Products (Paper) βž– Apple 2019
  4. Bootstrapping Conversational Agents with Weak Supervision (Paper) βž– IBM 2019

Generation πŸ”

  1. Better Language Models and Their Implications (Paper)OpenAI 2019
  2. Image GPT (Paper, Code) βž– OpenAI 2019
  3. Language Models are Few-Shot Learners (Paper) (GPT-3 Blog post) βž– OpenAI 2020
  4. Deep Learned Super Resolution for Feature Film Production (Paper) βž– Pixar 2020
  5. Unit Test Case Generation with Transformers βž– Microsoft 2021

Audio πŸ”

  1. Improving On-Device Speech Recognition with VoiceFilter-Lite (Paper)Google 2020
  2. The Machine Learning Behind Hum to Search βž– Google 2020

Validation and A/B Testing πŸ”

  1. Overlapping Experiment Infrastructure: More, Better, Faster Experimentation (Paper) βž– Google 2010
  2. The Reusable Holdout: Preserving Validity in Adaptive Data Analysis (Paper) βž– Google 2015
  3. Twitter Experimentation: Technical Overview βž– Twitter 2015
  4. It’s All A/Bout Testing: The Netflix Experimentation Platform βž– Netflix 2016
  5. Building Pinterest’s A/B Testing Platform βž– Pinterest 2016
  6. Experimenting to Solve Cramming βž– Twitter 2017
  7. Building an Intelligent Experimentation Platform with Uber Engineering βž– Uber 2017
  8. Scaling Airbnb’s Experimentation Platform βž– Airbnb 2017
  9. Meet Wasabi, an Open Source A/B Testing Platform (Code) βž– Intuit 2017
  10. Analyzing Experiment Outcomes: Beyond Average Treatment Effects βž– Uber 2018
  11. Under the Hood of Uber’s Experimentation Platform βž– Uber 2018
  12. Constrained Bayesian Optimization with Noisy Experiments (Paper) βž– Facebook 2018
  13. Reliable and Scalable Feature Toggles and A/B Testing SDK at Grab βž– Grab 2018
  14. Modeling Conversion Rates and Saving Millions Using Kaplan-Meier and Gamma Distributions (Code) βž– Better 2019
  15. Detecting Interference: An A/B Test of A/B Tests βž– LinkedIn 2019
  16. Announcing a New Framework for Designing Optimal Experiments with Pyro (Paper) (Paper) βž– Uber 2020
  17. Enabling 10x More Experiments with Traveloka Experiment Platform βž– Traveloka 2020
  18. Large Scale Experimentation at Stitch Fix (Paper) βž– Stitch Fix 2020
  19. Multi-Armed Bandits and the Stitch Fix Experimentation Platform βž– Stitch Fix 2020
  20. Experimentation with Resource Constraints βž– Stitch Fix 2020
  21. Computational Causal Inference at Netflix (Paper) βž– Netflix 2020
  22. Key Challenges with Quasi Experiments at Netflix βž– Netflix 2020
  23. Making the LinkedIn experimentation engine 20x faster βž– LinkedIn 2020
  24. Our Evolution Towards T-REX: The Prehistory of Experimentation Infrastructure at LinkedIn βž– LinkedIn 2020
  25. How to Use Quasi-experiments and Counterfactuals to Build Great Products βž– Shopify 2020
  26. Improving Experimental Power through Control Using Predictions as Covariate βž– DoorDash 2020
  27. Supporting Rapid Product Iteration with an Experimentation Analysis Platform βž– DoorDash 2020
  28. Improving Online Experiment Capacity by 4X with Parallelization and Increased Sensitivity βž– DoorDash 2020
  29. Leveraging Causal Modeling to Get More Value from Flat Experiment Results βž– DoorDash 2020
  30. Iterating Real-time Assignment Algorithms Through Experimentation βž– DoorDash 2020
  31. Spotify’s New Experimentation Platform (Part 1) (Part 2) βž– Spotify 2020
  32. Interpreting A/B Test Results: False Positives and Statistical Significance βž– Netflix 2021
  33. Interpreting A/B Test Results: False Negatives and Power βž– Netflix 2021
  34. Running Experiments with Google Adwords for Campaign Optimization βž– DoorDash 2021
  35. The 4 Principles DoorDash Used to Increase Its Logistics Experiment Capacity by 1000% βž– DoorDash 2021
  36. Experimentation Platform at Zalando: Part 1 - Evolution βž– Zalando 2021
  37. Designing Experimentation Guardrails βž– Airbnb 2021
  38. Network Experimentation at Scale(Paper] Facebook 2021
  39. Universal Holdout Groups at Disney Streaming βž– Disney 2021

Model Management πŸ”

  1. Operationalizing Machine Learningβ€”Managing Provenance from Raw Data to Predictions βž– Comcast 2018
  2. Overton: A Data System for Monitoring and Improving Machine-Learned Products (Paper) βž– Apple 2019
  3. Runway - Model Lifecycle Management at Netflix βž– Netflix 2020
  4. Managing ML Models @ Scale - Intuit’s ML Platform βž– Intuit 2020
  5. ML Model Monitoring - 9 Tips From the Trenches βž– Nubank 2021

Efficiency πŸ”

  1. GrokNet: Unified Computer Vision Model Trunk and Embeddings For Commerce (Paper) βž– Facebook 2020
  2. How We Scaled Bert To Serve 1+ Billion Daily Requests on CPUs βž– Roblox 2020
  3. Permute, Quantize, and Fine-tune: Efficient Compression of Neural Networks (Paper) βž– Uber 2021

Ethics πŸ”

  1. Building Inclusive Products Through A/B Testing (Paper) βž– LinkedIn 2020
  2. LiFT: A Scalable Framework for Measuring Fairness in ML Applications (Paper) βž– LinkedIn 2020

Infra πŸ”

  1. Reengineering Facebook AI’s Deep Learning Platforms for Interoperability βž– Facebook 2020
  2. Elastic Distributed Training with XGBoost on Ray βž– Uber 2021

MLOps Platforms πŸ”

  1. Meet Michelangelo: Uber’s Machine Learning Platform βž– Uber 2017
  2. Operationalizing Machine Learningβ€”Managing Provenance from Raw Data to Predictions βž– Comcast 2018
  3. Big Data Machine Learning Platform at Pinterest βž– Pinterest 2019
  4. Core Modeling at Instagram βž– Instagram 2019
  5. Open-Sourcing Metaflow - a Human-Centric Framework for Data Science βž– Netflix 2019
  6. Managing ML Models @ Scale - Intuit’s ML Platform βž– Intuit 2020
  7. Real-time Machine Learning Inference Platform at Zomato βž– Zomato 2020
  8. Introducing Flyte: Cloud Native Machine Learning and Data Processing Platform βž– Lyft 2020
  9. Building Flexible Ensemble ML Models with a Computational Graph βž– DoorDash 2021
  10. LyftLearn: ML Model Training Infrastructure built on Kubernetes βž– Lyft 2021
  11. β€œYou Don’t Need a Bigger Boat”: A Full Data Pipeline Built with Open-Source Tools (Paper) βž– Coveo 2021
  12. MLOps at GreenSteam: Shipping Machine Learning βž– GreenSteam 2021
  13. Evolving Reddit’s ML Model Deployment and Serving Architecture βž– Reddit 2021

Practices πŸ”

  1. Practical Recommendations for Gradient-Based Training of Deep Architectures (Paper) βž– Yoshua Bengio 2012
  2. Machine Learning: The High Interest Credit Card of Technical Debt (Paper) (Paper) βž– Google 2014
  3. Rules of Machine Learning: Best Practices for ML Engineering βž– Google 2018
  4. On Challenges in Machine Learning Model Management βž– Amazon 2018
  5. Machine Learning in Production: The Booking.com Approach βž– Booking 2019
  6. 150 Successful Machine Learning Models: 6 Lessons Learned at Booking.com (Paper) βž– Booking 2019
  7. Successes and Challenges in Adopting Machine Learning at Scale at a Global Bank βž– Rabobank 2019
  8. Challenges in Deploying Machine Learning: a Survey of Case Studies (Paper) βž– Cambridge 2020
  9. Reengineering Facebook AI’s Deep Learning Platforms for Interoperability βž– Facebook 2020
  10. The problem with AI developer tools for enterprises βž– Databricks 2020
  11. Continuous Integration and Deployment for Machine Learning Online Serving and Models βž– Uber 2021
  12. Tuning Model Performance βž– Uber 2021
  13. Maintaining Machine Learning Model Accuracy Through Monitoring βž– DoorDash 2021
  14. Building Scalable and Performant Marketing ML Systems at Wayfair βž– Wayfair 2021
  15. Our approach to building transparent and explainable AI systems βž– LinkedIn 2021
  16. 5 Steps for Building Machine Learning Models for Business βž– Shopify 2021

Team structure πŸ”

  1. Engineers Shouldn’t Write ETL: A Guide to Building a High Functioning Data Science Department βž– Stitch Fix 2016
  2. Building The Analytics Team At Wish βž– Wish 2018
  3. Beware the Data Science Pin Factory: The Power of the Full-Stack Data Science Generalist βž– Stitch Fix 2019
  4. Cultivating Algorithms: How We Grow Data Science at Stitch Fix βž– Stitch Fix
  5. Analytics at Netflix: Who We Are and What We Do βž– Netflix 2020
  6. Building a Data Team at a Mid-stage Startup: A Short Story βž– Erikbern 2021

Fails πŸ”

  1. When It Comes to Gorillas, Google Photos Remains Blind βž– Google 2010
  2. 160k+ High School Students Will Graduate Only If a Model Allows Them to βž– International Baccalaureate 2020
  3. An Algorithm That β€˜Predicts’ Criminality Based on a Face Sparks a Furor βž– Harrisburg University 2020
  4. It’s Hard to Generate Neural Text From GPT-3 About Muslims βž– OpenAI 2020
  5. A British AI Tool to Predict Violent Crime Is Too Flawed to Use βž– United Kingdom 2020
  6. More in awful-ai

References

[1] Machine Learning System Design

[2] Machine Learning Surveys 【forked from eugeneyan/ml-surveys】

[3] Applied Machine Learning 【forked from eugeneyan/applied-ml】