Applied ML
by eugeneyan · eugeneyan/applied-ml
Papers and blogs by organizations sharing their work on applied ML in production.
★ 27k stars525 projects
⑂ 0 forks0 open issues
Anomaly Detection
Deep Anomaly Detection with Spark and Tensorflowdatabricks.com
(Hopsworks Video) Swedbank, Hopsworks 2019
Detecting and Preventing Abuse on LinkedIn using Isolation Forestsengineering.linkedin.com
(Code) LinkedIn 2019
Improving the accuracy of our machine learning WAF using data augmentation and samplingblog.cloudflare.com
Cloudflare 2022
Audio
Classification
High-Precision Phrase-Based Document Classification on a Modern Scaleengineering.linkedin.com
(Paper) LinkedIn 2011
Uncovering Online Delivery Menu Best Practices with Machine Learningdoordash.engineering
DoorDash 2020
Using a Human-in-the-Loop to Overcome the Cold Start Problem in Menu Item Taggingdoordash.engineering
DoorDash 2020
Computer Vision
A Neural Weather Model for Eight-Hour Precipitation Forecastingai.googleblog.com
(Paper) Google 2020
How we Improved Computer Vision Metrics by More Than 5% Only by Cleaning Labelling Errorsdeepomatic.com
Deepomatic
Machine Learning-based Damage Assessment for Disaster Reliefai.googleblog.com
(Paper) Google 2020
Making machines recognize and transcribe conversations in meetings using audio and videomicrosoft.com
Microsoft 2019
Recognizing People in Photos Through Private On-Device Machine Learningmachinelearning.apple.com
Apple 2021
Data Discovery
Collect, Aggregate, and Visualize a Data Ecosystem's Metadatamarquezproject.github.io
(Code) WeWork
DataHub: A Generalized Metadata Search & Discovery Toolengineering.linkedin.com
(Code) LinkedIn 2019
How We Improved Data Discovery for Data Scientists at Spotifyengineering.atspotify.com
Spotify 2020
Metacat: Making Big Data Discoverable and Meaningful at Netflixnetflixtechblog.com
(Code) Netflix 2018
Using Amundsen to Support User Privacy via Metadata Collection at Squaredeveloper.squareup.com
Square 2020
Data Engineering
Building Scalable Real Time Event Processing with Kafka and Flinkdoordash.engineering
DoorDash 2022
Functions & DAGs: introducing Hamilton, a microframework for dataframe generationmultithreaded.stitchfix.com
Stitch Fix 2021
How DoorDash is Scaling its Data Platform to Delight Customers and Meet Growing Demanddoordash.engineering
DoorDash 2020
Introducing Fabricator: A Declarative Feature Engineering Frameworkdoordash.engineering
DoorDash 2022
LLessons Learned From Running Apache Airflow at Scaleshopifyengineering.myshopify.com
Shopify 2022
Unbundling Data Science Workflows with Metaflow and AWS Step Functionsnetflixtechblog.com
Netflix 2020
Understanding Data Storage and Ingestion for Large-Scale Deep Recommendation Model Trainingarxiv.org
Meta 2022
Data Quality
Efficiency
Embeddings
Ethics
A closer look at how LinkedIn integrates fairness into its AI productsengineering.linkedin.com
LinkedIn 2022
LiFT: A Scalable Framework for Measuring Fairness in ML Applicationsengineering.linkedin.com
(Paper) LinkedIn 2020
Fails
160k+ High School Students Will Graduate Only If a Model Allows Them topositivelysemidefinite.com
International Baccalaureate 2020
AFeature Stores
Accelerating Machine Learning with the Feature Store Servicetechnology.condenast.com
Condé Nast 2019
Building a Scalable ML Feature Store with Redis, Binary Serialization, and Compressiondoordash.engineering
DoorDash 2020
Building Riviera: A Declarative Real-Time Feature Engineering Frameworkdoordash.engineering
DoorDash 2021
Building the Activity Graph, Part 2 (Feature Storage Section)engineering.linkedin.com
LinkedIn 2017
Building the Model Behind DoorDash’s Expansive Merchant Selectiondoordash.engineering
DoorDash 2022
Developing scalable feature engineering DAGsouterbounds.com
Metaflow + Hamilton via Outerbounds 2022
Open sourcing Feathr – LinkedIn’s feature store for productive machine learningengineering.linkedin.com
LinkedIn 2022
Optimal Feature Discovery: Better, Leaner Machine Learning Models Through Information Theoryeng.uber.com
Uber 2021
Rapid Experimentation Through Standardization: Typed AI features for LinkedIn’s Feedengineering.linkedin.com
LinkedIn 2020
Forecasting
Greykite: A flexible, intuitive, and fast forecasting libraryengineering.linkedin.com
LinkedIn 2021
Generation
Graph
Information Extraction
Infra
MLOps Platforms
"Building Flexible Ensemble ML Models with a Computational Graphdoordash.engineering
DoorDash 2021
CDARWIN: Data Science and Artificial Intelligence Workbench at LinkedInengineering.linkedin.com
LinkedIn 2022
Deployment for Free - A Machine Learning Platform for Stitch Fix's Data Scientistsmultithreaded.stitchfix.com
Stitch Fix 2022
Open-Sourcing Metaflow - a Human-Centric Framework for Data Sciencenetflixtechblog.com
Netflix 2019
TModel Management
Dealing with Train-serve Skew in Real-time ML Models: A Short Guidebuilding.nubank.com.br
Nubank 2023
Natural Language Processing
Deep Learning to Translate Between Programming Languagesai.facebook.com
(Paper, Code) Facebook 2020
DeText: A deep NLP Framework for Intelligent Text Understandingengineering.linkedin.com
(Code) LinkedIn 2020
GeDi: A Powerful New Method for Controlling Language Modelsblog.einstein.ai
(Paper, Code) Salesforce 2020
Give Me Jeans not Shoes: How BERT Helps Us Deliver What Clients Wantmultithreaded.stitchfix.com
Stitch Fix 2019
GGoal-Oriented End-to-End Conversational Models with Profile Features in a Real-World Settingamazon.science
(Paper) Amazon 2019
How Natural Language Processing Helps LinkedIn Members Get Support Easilyengineering.linkedin.com
LinkedIn 2019
PEGASUS: A State-of-the-Art Model for Abstractive Text Summarizationai.googleblog.com
(Paper, Code) Google 2020
Textless NLP: Generating expressive speech from raw audioai.facebook.com
(Part 1) (Part 2) (Part 3) (Code and Pretrained Models) Facebook 2021
The State-of-the-art Open-Domain Chatbot in Chinese and Englishresearch.baidu.com
(Paper) Baidu 2020
Optimization
The Data and Science behind GrabShare Carpoolingieeexplore.ieee.org
(Part 1) (PAPER NEEDED) Grab 2017
Practices
Automatic Retraining for Machine Learning Models: Tips and Lessons Learnedbuilding.nubank.com.br
Nubank 2022
Building and Maintaining Internal Tools for DS/ML teams: Lessons Learnedbuilding.nubank.com.br
Nubank 2024
DData Is An Art, Not Just A Science—And Storytelling Is The Keyshopifyengineering.myshopify.com
Shopify 2022
Our approach to building transparent and explainable AI systemsengineering.linkedin.com
LinkedIn 2021
Privacy-preserving Machine Learning
Federated Learning with Formal Differential Privacy Guaranteesai.googleblog.com
(Paper) Google 2022
Federated Learning: Collaborative Machine Learning without Centralized Training Dataai.googleblog.com
(Paper) Google 2017
MPC-based machine learning: Achieving end-to-end privacy-preserving machine learningresearch.facebook.com
(Paper) Facebook 2022
Recommendation
A Closer Look at the AI Behind Course Recommendations on LinkedIn Learning (Part 1)engineering.linkedin.com
LinkedIn 2020
A Closer Look at the AI Behind Course Recommendations on LinkedIn Learning (Part 2)engineering.linkedin.com
LinkedIn 2020
Amazon.com Recommendations: Item-to-Item Collaborative Filteringieeexplore.ieee.org
(Paper) Amazon 2003
Beyond Matrix Factorization: Using hybrid features for user-business recommendationsengineeringblog.yelp.com
Yelp 2022
Building a Heterogeneous Social Network Recommendation Systemengineering.linkedin.com
LinkedIn 2020
DFor Your Ears Only: Personalizing Spotify Home with Machine Learningengineering.atspotify.com
Spotify 2020
How 20th Century Fox uses ML to predict a movie audiencecloud.google.com
(Paper) 20th Century Fox 2018
How Pinterest Leverages Realtime User Actions in Recommendation to Boost Homefeed Engagement Volumemedium.com
Pinterest 2022
Improving job matching with machine-learned activity featuresengineering.linkedin.com
LinkedIn 2022
Personalized Cuisine Filter Based on Customer Preference and Local Popularitydoordash.engineering
DoorDash 2020
Reach for the Top: How Spotify Built Shortcuts in Just Six Monthsengineering.atspotify.com
Spotify 2020
Recommend API: Unified end-to-end machine learning infrastructure to generate recommendationsslack.engineering
Slack 2022
RecSysOps: Best Practices for Operating a Large-Scale Recommender Systemnetflixtechblog.medium.com
Netflix 2022
Talent Search and Recommendation Systems at LinkedIn: Practical Challenges and Lessons Learnedarxiv.org
(Paper) LinkedIn 2018
TThe Evolution of Kit: Automating Marketing Using Machine Learningengineering.shopify.com
Shopify 2020
To Be Continued: Helping you find shows to continue watching on Netflixnetflixtechblog.com
Netflix 2016
Understanding Data Storage and Ingestion for Large-Scale Deep Recommendation Model Trainingarxiv.org
(Paper) Meta 2021
Regression
Solving for Unobserved Data in a Regression Model Using a Simple Data Adjustmentdoordash.engineering
DoorDash 2020
Reinforcement Learning
Bandits for Online Calibration: An Application to Content Moderation on Social Media Platformsarxiv.org
Meta 2022
Selecting the Best Image for Each Merchant Using Exploration and Machine Learningdoordash.engineering
DoorDash 2023
Shifting Consumption towards Diverse content via Reinforcement Learningresearch.atspotify.com
(Paper) Spotify 2022
Search & Ranking
MOBIUS: Towards the Next Generation of Query-Ad Matching in Baidu’s Sponsored Searchresearch.baidu.com
Baidu 2019
Quality Matches Via Personalized AI for Hirer and Seeker Preferencesengineering.linkedin.com
LinkedIn 2020
Query Understanding for Surfacing Under-served Music Contentresearch.atspotify.com
(Paper) Spotify 2020
Stitching together spaces for query-based recommendationsmultithreaded.stitchfix.com
Stitch Fix 2021
The AI Behind LinkedIn Recruiter Search and recommendation systemsengineering.linkedin.com
LinkedIn 2019
Things Not Strings: Understanding Search Intent with Better Recalldoordash.engineering
DoorDash 2020
Sequence Modelling
Using deep learning to detect abusive sequences of member activityengineering.linkedin.com
(Video) LinkedIn 2021
Team structure
Beware the Data Science Pin Factory: The Power of the Full-Stack Data Science Generalistmultithreaded.stitchfix.com
Stitch Fix 2019
Cultivating Algorithms: How We Grow Data Science at Stitch Fixcultivating-algos.stitchfix.com
Stitch Fix
Data Scientist x Machine Learning Engineer Roles: How are they different? How are they alike?building.nubank.com.br
Nubank 2022
Engineers Shouldn’t Write ETL: A Guide to Building a High Functioning Data Science Departmentmultithreaded.stitchfix.com
Stitch Fix 2016
What is the most effective way to structure a data science team?towardsdatascience.com
Udemy 2017
Validation and A/B Testing
HHow to Use Quasi-experiments and Counterfactuals to Build Great Productsengineering.shopify.com
Shopify 2020
Improving Experimental Power through Control Using Predictions as Covariatedoordash.engineering
DoorDash 2020
Improving Online Experiment Capacity by 4X with Parallelization and Increased Sensitivitydoordash.engineering
DoorDash 2020
Interpreting A/B Test Results: False Positives and Statistical Significancenetflixtechblog.com
Netflix 2021
Iterating Real-time Assignment Algorithms Through Experimentationdoordash.engineering
DoorDash 2020
Leveraging Causal Modeling to Get More Value from Flat Experiment Resultsdoordash.engineering
DoorDash 2020
Meet Dash-AB — The Statistics Engine of Experimentation at DoorDashdoordash.engineering
DoorDash 2022
Modeling Conversion Rates and Saving Millions Using Kaplan-Meier and Gamma Distributionsbetter.engineering
(Code) Better 2019
Multi-Armed Bandits and the Stitch Fix Experimentation Platformmultithreaded.stitchfix.com
Stitch Fix 2020
Our Evolution Towards T-REX: The Prehistory of Experimentation Infrastructure at LinkedInengineering.linkedin.com
LinkedIn 2020
Running Experiments with Google Adwords for Campaign Optimizationdoordash.engineering
DoorDash 2021
Supporting Rapid Product Iteration with an Experimentation Analysis Platformdoordash.engineering
DoorDash 2020
The 4 Principles DoorDash Used to Increase Its Logistics Experiment Capacity by 1000%doordash.engineering
DoorDash 2021
The Reusable Holdout: Preserving Validity in Adaptive Data Analysisai.googleblog.com
(Paper) Google 2015
Weak Supervision