I build recommendation systems and AI-powered experiences that serve customers at scale. Currently, I'm a Principal Applied Scientist at Amazon. Outside of work, I also...

• Write,speak, andprototype on ideas in machine learning, RecSys, and LLMs.
• Teach how to apply ML/LLMs effectively atApplyingML.com &Applied-LLMs.org.
• Send a newsletter about data, ML, and what I'm learning to 10,000+ subscribers.
• In 2025 I'm learning:RecSys × LLMs,semantic IDs,long-context Q&A, and agents.
• Fun fact: I don't use the QWERTY keyboard (I useDvorak instead).

• 2025 Year in Review - Sun, 14 Dec 2025
• Product Evals in Three Simple Steps - Sun, 23 Nov 2025
• Advice for New Principal Tech ICs (i.e., Notes to Myself) - Sun, 19 Oct 2025
• Training an LLM-RecSys Hybrid for Steerable Recs with Semantic IDs - Sun, 14 Sep 2025
• Evaluating Long-Context Question & Answer Systems - Sun, 22 Jun 2025

View the archives (209 posts) @eugeneyan.com.

PinnedLoading

1. applied-mlapplied-mlPublic

   📚 Papers & tech blogs by companies sharing their work on data science & machine learning in production.

   28.7k 3.8k

2. ml-design-docsml-design-docsPublic

   📝 Design doc template & examples for machine learning systems (requirements, methodology, implementation, etc.)

   644 109

3. obsidian-copilotobsidian-copilotPublic

   🤖 A prototype assistant for writing and thinking

   Python 557 39

4. news-agentsnews-agentsPublic

   📰 Building News Agents to Summarize News with MCP, Q, and tmux

   Python 312 35

5. testing-mltesting-mlPublic

   🔍 Minimal examples of machine learning tests for implementation, behaviour, and performance.

   Python 265 54

6. semantic-ids-llmsemantic-ids-llmPublic

   Semantic IDs: How to train an LLM-Recommender Hybrid with steerability and reasoning on recommendations.

   Jupyter Notebook 96 25