applied-ml

Curated papers, articles, and blogs on data science & machine learning in production. ⚙️

Figuring out how to implement your ML project? Learn how other organizations did it:

• How the problem is framed 🔎(e.g., personalization as recsys vs. search vs. sequences)
• What machine learning techniques worked ✅ (and sometimes, what didn't ❌)
• Why it works, the science behind it with research, literature, and references 📂
• What real-world results were achieved (so you can better assess ROI ⏰💰📈)

P.S., Want a summary of ML advancements? 👉ml-surveys

Table of Contents

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. Practices
27. Team Structure
28. Fails

Data Quality

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

Data Engineering

1. Zipline: Airbnb’s Machine Learning Data Management Platform Airbnb
2. Sputnik: Airbnb’s Apache Spark Framework for Data Engineering Airbnb
3. Unbundling Data Science Workflows with Metaflow and AWS Step Functions Netflix
4. How DoorDash is Scaling its Data Platform to Delight Customers and Meet Growing Demand DoorDash
5. Revolutionizing Money Movements at Scale with Strong Data Consistency Uber

Data Discovery

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

Feature Stores

1. Introducing Feast: an open source feature store for machine learning (Code)Gojek
2. Feast: Bridging ML Models and Data Gojek
3. Building a Scalable ML Feature Store with Redis, Binary Serialization, and Compression DoorDash
4. Michelangelo Palette: A Feature Engineering Platform at Uber Uber
5. Distributed Time Travel for Feature Generation Netflix
6. The Architecture That Powers Twitter's Feature Store Twitter
7. Rapid Experimentation Through Standardization: Typed AI features for LinkedIn’s Feed LinkedIn

Classification

1. High-Precision Phrase-Based Document Classification on a Modern Scale (Paper) LinkedIn
2. Chimera: Large-scale Classification using Machine Learning, Rules, and Crowdsourcing (Paper) WalmartLabs
3. Large-scale Item Categorization for e-Commerce (Paper) DianPing, eBay
4. Large-scale Item Categorization in e-Commerce Using Multiple Recurrent Neural Networks (Paper) NAVER
5. Categorizing Products at Scale Shopify
6. Learning to Diagnose with LSTM Recurrent Neural Networks (Paper) Google
7. Discovering and Classifying In-app Message Intent at Airbnb Airbnb
8. How We Built the Good First Issues Feature GitHub
9. Teaching Machines to Triage Firefox Bugs Mozilla
10. Testing Firefox More Efficiently with Machine Learning Mozilla
11. Using ML to Subtype Patients Receiving Digital Mental Health Interventions (Paper) Microsoft
12. Prediction of Advertiser Churn for Google AdWords (Paper) Google
13. Scalable Data Classification for Security and Privacy (Paper) Facebook

Regression

1. Using Machine Learning to Predict Value of Homes On Airbnb Airbnb
2. Using Machine Learning to Predict the Value of Ad Requests Twitter
3. Open-Sourcing Riskquant, a Library for Quantifying Risk (Code) NetFlix

Forecasting

1. Forecasting at Uber: An Introduction Uber
2. Engineering Extreme Event Forecasting at Uber with RNN Uber
3. Transforming Financial Forecasting with Data Science and Machine Learning at Uber Uber
4. Under the Hood of Gojek’s Automated Forecasting Tool GoJek
5. BusTr: Predicting Bus Travel Times from Real-Time Traffic (Paper, Video) Google
6. Retraining Machine Learning Models in the Wake of COVID-19 DoorDash
7. Automatic Forecasting using Prophet, Databricks, Delta Lake and MLflow (Paper, Code) Atlassian

Recommendation

1. Amazon.com Recommendations: Item-to-Item Collaborative Filtering (Paper) Amazon
2. Temporal-Contextual Recommendation in Real-Time (Paper) Amazon
3. P-Companion: A Framework for Diversified Complementary Product Recommendation (Paper) Amazon
4. Recommending Complementary Products in E-Commerce Push Notifications (Paper) Alibaba
5. Behavior Sequence Transformer for E-commerce Recommendation in Alibaba (Paper) Alibaba
6. TPG-DNN: A Method for User Intent Prediction with Multi-task Learning (Paper) Alibaba
7. PURS: Personalized Unexpected Recommender System for Improving User Satisfaction (Paper) Alibaba
8. Session-based Recommendations with Recurrent Neural Networks (Paper) Telefonica
9. How 20th Century Fox uses ML to predict a movie audience (Paper) 20th Century Fox
10. Deep Neural Networks for YouTube Recommendations YouTube
11. Personalized Recommendations for Experiences Using Deep Learning TripAdvisor
12. E-commerce in Your Inbox: Product Recommendations at Scale Yahoo
13. Product Recommendations at Scale (Paper) Yahoo
14. Powered by AI: Instagram’s Explore recommender system Facebook
15. Netflix Recommendations: Beyond the 5 stars (Part 1 (Part 2) Netflix
16. Learning a Personalized Homepage Netflix
17. Artwork Personalization at Netflix Netflix
18. To Be Continued: Helping you find shows to continue watching on Netflix Netflix
19. Calibrated Recommendations (Paper) Netflix
20. Food Discovery with Uber Eats: Recommending for the Marketplace Uber
21. Food Discovery with Uber Eats: Using Graph Learning to Power Recommendations Uber
22. How Music Recommendation Works — And Doesn’t Work Spotify
23. Music recommendation at Spotify Spotify
24. Recommending Music on Spotify with Deep Learning Spotify
25. For Your Ears Only: Personalizing Spotify Home with Machine Learning Spotify
26. Reach for the Top: How Spotify Built Shortcuts in Just Six Months Spotify
27. Explore, Exploit, and Explain: Personalizing Explainable Recommendations with Bandits (Paper) Spotify
28. Contextual and Sequential User Embeddings for Large-Scale Music Recommendation (Paper) Spotify
29. The Evolution of Kit: Automating Marketing Using Machine Learning Shopify
30. Using Machine Learning to Predict what File you Need Next (Part 1) Dropbox
31. Using Machine Learning to Predict what File you Need Next (Part 2) Dropbox
32. Personalized Recommendations in LinkedIn Learning LinkedIn
33. A Closer Look at the AI Behind Course Recommendations on LinkedIn Learning (Part 1) LinkedIn
34. A Closer Look at the AI Behind Course Recommendations on LinkedIn Learning (Part 2) LinkedIn
35. Learning to be Relevant: Evolution of a Course Recommendation System (PAPER NEEDED)LinkedIn
36. Building a Heterogeneous Social Network Recommendation System LinkedIn
37. How TikTok recommends videos #ForYou ByteDance
38. A Meta-Learning Perspective on Cold-Start Recommendations for Items (Paper) Twitter
39. Zero-Shot Heterogeneous Transfer Learning from RecSys to Cold-Start Search Retrieval (Paper) Google
40. Improved Deep & Cross Network for Feature Cross Learning in Web-scale LTR Systems (Paper) Google
41. Personalized Channel Recommendations in Slack Slack
42. Deep Retrieval: End-to-End Learnable Structure Model for Large-Scale Recommendations (Paper) ByteDance
43. Future Data Helps Training: Modeling Future Contexts for Session-based Recommendation (Paper) Tencent
44. Using AI to Help Health Experts Address the COVID-19 Pandemic Facebook
45. A Case Study of Session-based Recommendations in the Home-improvement Domain (Paper) Home Depot
46. Balancing Relevance and Discovery to Inspire Customers in the IKEA App (Paper) Ikea
47. How we use AutoML, Multi-task learning and Multi-tower models for Pinterest Ads Pinterest

Search & Ranking

1. Amazon Search: The Joy of Ranking Products (Paper, Video, Code) Amazon
2. Why Do People Buy Seemingly Irrelevant Items in Voice Product Search? (Paper) Amazon
3. How Lazada Ranks Products to Improve Customer Experience and Conversion Lazada
4. Using Deep Learning at Scale in Twitter’s Timelines Twitter
5. Machine Learning-Powered Search Ranking of Airbnb Experiences Airbnb
6. Applying Deep Learning To Airbnb Search (Paper) Airbnb
7. Managing Diversity in Airbnb Search (Paper) Airbnb
8. Improving Deep Learning for Airbnb Search (Paper) Airbnb
9. Ranking Relevance in Yahoo Search (Paper) Yahoo
10. An Ensemble-based Approach to Click-Through Rate Prediction for Promoted Listings at Etsy (Paper) Etsy
11. Learning to Rank Personalized Search Results in Professional Networks (Paper) LinkedIn
12. Entity Personalized Talent Search Models with Tree Interaction Features (Paper) LinkedIn
13. In-session Personalization for Talent Search (Paper) LinkedIn
14. The AI Behind LinkedIn Recruiter search and recommendation systems LinkedIn
15. Quality Matches Via Personalized AI for Hirer and Seeker Preferences LinkedIn
16. Understanding Dwell Time to Improve LinkedIn Feed Ranking LinkedIn
17. Ads Allocation in Feed via Constrained Optimization (Paper, Video) LinkedIn
18. AI at Scale in Bing Microsoft
19. Query Understanding Engine in Traveloka Universal Search Traveloka
20. The Secret Sauce Behind Search Personalisation GoJek
21. Food Discovery with Uber Eats: Building a Query Understanding Engine Uber
22. Neural Code Search: ML-based Code Search Using Natural Language Queries Facebook
23. Bayesian Product Ranking at Wayfair Wayfair
24. COLD: Towards the Next Generation of Pre-Ranking System (Paper) Alibaba
25. Understanding Searches Better Than Ever Before (Paper) Google
26. Shop The Look: Building a Large Scale Visual Shopping System at Pinterest (Paper, Video) Pinterest
27. GDMix: A Deep Ranking Personalization Framework (Code) LinkedIn
28. Bringing Personalized Search to Etsy Etsy
29. Building a Better Search Engine for Semantic Scholar Allen Institute for AI

Embeddings

1. Billion-scale Commodity Embedding for E-commerce Recommendation in Alibaba (Paper) Alibaba
2. Embeddings@Twitter Twitter
3. Listing Embeddings in Search Ranking (Paper) Airbnb
4. Understanding Latent Style Stitch Fix
5. Towards Deep and Representation Learning for Talent Search at LinkedIn (Paper) LinkedIn
6. Vector Representation Of Items, Customer And Cart To Build A Recommendation System (Paper) Sears
7. Machine Learning for a Better Developer Experience Netflix
8. Announcing ScaNN: Efficient Vector Similarity Search (Paper, Code) Google

Natural Language Processing

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

Sequence Modelling

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

Computer Vision

1. Categorizing Listing Photos at Airbnb Airbnb
2. Amenity Detection and Beyond — New Frontiers of Computer Vision at Airbnb Airbnb
3. Powered by AI: Advancing product understanding and building new shopping experiences Facebook
4. Creating a Modern OCR Pipeline Using Computer Vision and Deep Learning Dropbox
5. How we Improved Computer Vision Metrics by More Than 5% Only by Cleaning Labelling Errors Deepomatic
6. A Neural Weather Model for Eight-Hour Precipitation Forecasting (Paper) Google
7. Machine Learning-based Damage Assessment for Disaster Relief (Paper) Google
8. RepNet: Counting Repetitions in Videos (Paper) Google
9. Converting Text to Images for Product Discovery (Paper) Amazon
10. How Disney Uses PyTorch for Animated Character Recognition Disney
11. Image Captioning as an Assistive Technology (Video) IBM
12. AI for AG: Production machine learning for agriculture Blue River
13. AI for Full-Self Driving at Tesla Tesla
14. On-device Supermarket Product Recognition Google
15. Using Machine Learning to Detect Deficient Coverage in Colonoscopy Screenings (Paper) Google
16. Shop The Look: Building a Large Scale Visual Shopping System at Pinterest (Paper, Video) Pinterest
17. Developing Real-Time, Automatic Sign Language Detection for Video Conferencing (Paper) Google

Reinforcement Learning

1. Deep Reinforcement Learning for Sponsored Search Real-time Bidding (Paper) Alibaba
2. Dynamic Pricing on E-commerce Platform with Deep Reinforcement Learning (Paper) Alibaba
3. Budget Constrained Bidding by Model-free Reinforcement Learning in Display Advertising (Paper) Alibaba
4. Productionizing Deep Reinforcement Learning with Spark and MLflow Zynga
5. Deep Reinforcement Learning in Production Part1 Part 2 Zynga
6. Building AI Trading Systems Denny Britz

Anomaly Detection

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

Graph

1. Building The LinkedIn Knowledge Graph LinkedIn
2. Retail Graph — Walmart’s Product Knowledge Graph Walmart
3. Food Discovery with Uber Eats: Using Graph Learning to Power Recommendations Uber
4. AliGraph: A Comprehensive Graph Neural Network Platform (Paper) Alibaba
5. Scaling Knowledge Access and Retrieval at Airbnb Airbnb
6. Traffic Prediction with Advanced Graph Neural Networks DeepMind
7. SimClusters: Community-Based Representations for Recommendations (Paper, Video) Twitter

Optimization

1. How Trip Inferences and Machine Learning Optimize Delivery Times on Uber Eats Uber
2. Next-Generation Optimization for Dasher Dispatch at DoorDash DoorDash
3. Matchmaking in Lyft Line (Part 1) (Part 2) (Part 3) Lyft
4. The Data and Science behind GrabShare Carpooling (PAPER NEEDED) Grab
5. Optimization of Passengers Waiting Time in Elevators Using Machine Learning Thyssen Krupp AG
6. Think out of the package: Recommending package types for e-commerce shipments (Paper) Amazon

Information Extraction

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

Weak Supervision

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

Generation

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

Audio

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

Validation and A/B Testing

1. The Reusable Holdout: Preserving Validity in Adaptive Data Analysis (Paper) Google
2. Detecting Interference: An A/B Test of A/B Tests LinkedIn
3. Experimenting to Solve Cramming Twitter
4. Announcing a New Framework for Designing Optimal Experiments with Pyro (Paper) (Paper) Uber
5. Enabling 10x More Experiments with Traveloka Experiment Platform Traveloka
6. Large Scale Experimentation at Stitch Fix (Paper) Stitch Fix
7. Multi-Armed Bandits and the Stitch Fix Experimentation Platform Stitch Fix
8. Experimentation with Resource Constraints Stitch Fix
9. Modeling Conversion Rates and Saving Millions Using Kaplan-Meier and Gamma Distributions (Code) Better
10. Computational Causal Inference at Netflix (Paper) Netflix
11. Key Challenges with Quasi Experiments at Netflix Netflix
12. Constrained Bayesian Optimization with Noisy Experiments (Paper) Facebook
13. Supporting Rapid Product Iteration with an Experimentation Analysis Platform Curie
14. Our Evolution Towards T-REX: The Prehistory of Experimentation Infrastructure at LinkedIn LinkedIn
15. How to Use Quasi-experiments and Counterfactuals to Build Great Products Shopify
16. Improving Online Experiment Capacity by 4X with Parallelization and Increased Sensitivity DoorDash
17. Spotify’s New Experimentation Platform (Part 1) (Part 2) Spotify
18. Overlapping Experiment Infrastructure: More, Better, Faster Experimentation (Paper) Google

Model Management

1. Runway - Model Lifecycle Management at Netflix Netflix

Efficiency

1. GrokNet: Unified Computer Vision Model Trunk and Embeddings For Commerce (Paper) Facebook
2. Permute, Quantize, and Fine-tune: Efficient Compression of Neural Networks (Paper) Uber

Ethics

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

Practices

1. Practical Recommendations for Gradient-Based Training of Deep Architectures (Paper) Yoshua Bengio
2. Machine Learning: The High Interest Credit Card of Technical Debt (Paper) (Paper) Google
3. Rules of Machine Learning: Best Practices for ML Engineering Google
4. On Challenges in Machine Learning Model Management Amazon
5. Machine Learning in Production: The Booking.com Approach Booking
6. 150 Successful Machine Learning Models: 6 Lessons Learned at Booking.com (Paper) Booking
7. Successes and Challenges in Adopting Machine Learning at Scale at a Global Bank Rabobank

Team structure

1. Engineers Shouldn’t Write ETL: A Guide to Building a High Functioning Data Science Department Stitch Fix
2. Beware the Data Science Pin Factory: The Power of the Full-Stack Data Science Generalist Stitch Fix
3. Cultivating Algorithms: How We Grow Data Science at Stitch Fix StitchFix
4. Analytics at Netflix: Who We Are and What We Do Netflix

Fails

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

P.S., Want a summary of ML advancements? Get up to speed with survey papers 👉ml-surveys