Hi, I'm Eugene Yan.
I work to bridge the field and the frontier, and help build safe, reliable AI systems that scale. Currently, I'm a Member of Technical Staff at Anthropic.
I write and speak about AI-powered products, recommendation systems, and engineering, and share prototypes here.
Latest
All 182 posts →- Using LLMs to Secure Source Code
AI Engineer 2026 · Sharing what we learned from Project Glasswing.
- Patterns for Building Cybersecurity Evals
A sandboxed target, inputs that influence task difficulty, tools, and a grader.
- Using LLMs to Secure Source Code
Build a threat model, discover vulnerabilities, verify, triage, and patch.
- How to Work and Compound with AI
Context as infra, taste as config, verify for autonomy, scale via delegation, close the loop.
- 2025 Year in Review
An eventful year of progress in health and career, while making time for travel and reflection.
Reader favourites
All 21 →- Improving RecSys & Search with LLMsModel architecture, data generation, etc.
- LLM-EvaluatorsUse cases, techniques, alignment, and critiques
- How to Interview ML/AI EngineersHow to run phone screens, loops, and debriefs
- Prompting FundamentalsStructured I/O, prefilling, n-shot, CoT, etc.
- A Year of Building with LLMsTactical, operational, and strategic lessons
Explore by theme
Start here →- ML systems in industryHow LLM, RecSys, and search systems are built
- ML techniquesSurveys from prompting and evals to bandits
- ML & engineeringDesign patterns, testing, and ML in production
- Mechanisms for teamsPractices for business, product, and tech teams
- Learning & careerGrowing as an engineer, scientist, and leader
Featured talks
All 32 talks →- What We Learned from a Year of LLMsAI Engineer 2024 Keynote
- LLMs for Recommendation ExperiencesNetflix PRS 2024
- Building Blocks for LLM SystemsAI Engineer 2023 Keynote
- Is the Juice Worth the Squeeze?RecSys 2022 Keynote
- System Design for RecSys & SearchMLOps Community
Prototypes
All prototypes →- Semantic IDsTraining an LLM-RecSys hybrid for steerable recs
- News AgentsAutomating my daily news flash with agents
- AI Reading ClubPrototyping an AI-powered reading experience
- AlignEvalMaking evals easy, fun, and semi-automated
- AI CoachTalk to Tara, an AI coach, anytime
182 posts, 32 talks, 399,172 words, and countless hours.