About
I write about the practical side of building and evaluating AI systems.
I am Jing Lu, and this site is where I collect technical notes on LLM systems, AI agents, post-training, RLHF, evaluation, recommender systems, and AI product design.
The common thread is implementation: how the research idea becomes a working system, where it breaks, and what engineering patterns make it reliable enough to ship.
Areas I Write About
- LLM agents, tool use, and agent runtime design
- Agent and model evaluation
- RLHF, post-training, and preference optimization
- Instruction following and reasoning systems
- Generative UI and AI product interfaces