How to use applied-ml: a guide to ML in production papers

What applied-ml is and who it is for

applied-ml is a curated list, not a software tool. It collects links to material on data science and machine learning in production, and the README says it is meant to help you learn how other organizations did it.

It suits engineers planning an ML project, data scientists looking for real-world case studies on recommendation, search or data quality, students and researchers who want practical context next to academic papers, and team leads comparing infrastructure choices such as feature stores or data pipelines.

• How the problem is framed, for example personalization as recsys vs. search vs. sequences
• What machine learning techniques worked, and sometimes what did not
• Why it works, with research, literature and references
• What real-world results were reported, which helps you judge ROI

Get the list: read online or clone it to read offline

You do not need to install anything. The entries are links, so you can read the README directly on GitHub.

If you want a local copy to read offline, the README gives this command. The source lists no other prerequisites.

Find entries by topic using the table of contents

The README opens with a numbered table of contents that links to each topic. Topics include Data Quality, Data Engineering, Data Discovery, Feature Stores, Classification, Regression, Forecasting, Recommendation, Search & Ranking, Embeddings, Natural Language Processing, Computer Vision, Reinforcement Learning, Anomaly Detection, Validation and A/B Testing, Model Management, MLOps Platforms, Team Structure and Fails, among others.

If you are choosing infrastructure, start with Data Engineering, Feature Stores and Infra. If you are working on a product problem, start with the topic that matches it, such as Recommendation or Search & Ranking.

• Data Quality: monitoring and validating data that feeds models
• Feature Stores: where teams manage features for models
• Recommendation and Search & Ranking: real-world case studies
• MLOps Platforms and Model Management: how teams run models after launch

Read each entry for the company, year and kind of source

Each entry is tagged with the company that wrote it and the year it was published, shown as tags after the link. Some entries also link to a paper, a code repository or a talk, shown in parentheses.

Entries are a mix of engineering blog posts, research papers, conference talks and project pages. Talks are often linked to video, so check the link type before you plan your reading time.

• Company tag: who did the work
• Year tag: when it was published, useful for judging how current it is
• Extra links: papers, code or videos when the README provides them

Use applied-ml alongside other reading lists

applied-ml is one of several lists. The README points to ml-surveys for summaries of ML advancements and to applyingML for guides and interviews on applying ML. Comparable projects named in the analysis include awesome-production-machine-learning and awesome-machine-learning.

Use applied-ml for company case studies and combine it with these lists when you need background or a different angle.

• ml-surveys: summaries of ML advancements
• applyingML: guides and interviews on applying ML
• awesome-production-machine-learning and awesome-machine-learning: other curated lists

Check currency before you rely on an entry

The analysis notes that the repository has not been updated recently and that the visible entries are older. Treat it as a historical view of how companies approached problems, not a guide to current tools.

Compare the year tag with the problem you are solving. Older entries can still show useful framing, trade-offs and reported results, but check newer sources before choosing a specific technique.