What Topics Does Eugene Yan Write About on eugeneyan.com?

Eugene Yan writes on eugeneyan.com about recommendation systems (RecSys) and search, building and evaluating LLM/AI systems and products, ML/AI engineering practice and career development, and personal updates such as prototype projects and annual reviews. The site is a personal blog and newsletter by Eugene Yan, who describes himself as a Member of Technical Staff at Anthropic and says he has led ML/AI teams at Amazon, Alibaba, Lazada, and a healthtech Series A company. If you want practical, engineering-oriented writing on applied AI and RecSys rather than general news, this site is worth following.

Core Topics

Recommendation Systems and Search

RecSys is a long-running theme. Posts cover improving recommendation and search systems with LLM techniques, training hybrid LLM-RecSys models with semantic IDs for steerable recommendations, and intuition behind attention and the Transformer. The site also links to ApplyingML.com, where Yan teaches how to apply ML/LLMs effectively.

LLM/AI Systems and Products

This is the largest cluster. Representative posts include:

• Patterns for Building LLM-based Systems & Products — architectural and engineering patterns for LLM applications.
• What We've Learned From A Year of Building with LLMs — lessons from production LLM work.
• Evaluating the Effectiveness of LLM-Evaluators (aka LLM-as-Judge) — how to assess LLM judges.
• Product Evals in Three Simple Steps and Evaluating Long-Context Question & Answer Systems — practical evaluation methods.
• Prompting Fundamentals and How to Apply them Effectively — prompting basics and application.
• Using LLMs to Secure Source Code and Patterns for Building Cybersecurity Evals — security-focused LLM applications.

ML/AI Engineering Practice and Career

Posts such as How to Interview and Hire ML/AI Engineers and Advice for New Principal Tech ICs (i.e., Notes to Myself) address hiring and career growth for technical ICs. How to Work and Compound with AI covers working effectively with AI tools.

Prototypes and Personal Sharing

Yan shares prototype apps including AI Reading Club, AlignEval, Obsidian Copilot, and AI Coach, plus annual reviews such as 2025 Year in Review. These give a sense of how the ideas are applied in practice.

Format and Access

The site lists a subscribe page for email updates. Pricing or access restrictions are not stated in the available material, so check the subscribe page directly if you need those details.

Who It Suits

• ML/AI engineers who want concrete patterns, evaluation methods, and production lessons for LLM systems.
• RecSys and search practitioners interested in combining LLMs with recommendation systems.
• Engineers preparing for senior IC roles or hiring who want interview and career notes.
• Readers who prefer engineering depth over news — the writing is framed around building safe, reliable AI systems that scale.

If your interest is broad AI industry news or beginner tutorials, the site's focus on applied engineering and evaluation may be narrower than you need.