Hello, I'm
Christine Hu
Building agentic RL for video AI.
About
Builder, explorer, wanderer.
Currently building in a nomad way. You'll find me in SF, the Dolomites, Yosemite, and maybe a Kyoto temple. Landscape photography, mushroom foraging, Japanese cuisine, running trails. Practitioner of walking meditation, snow mountain enthusiast, backcountry skier.
Previously: Applied Scientist & Tech Lead at Microsoft AI, focused on long-horizon multimodal reasoning. Built Evoke (ICLR '24) — automatic prompt engineering that evolved into automatic skills engineering and automatic harness engineering. Published TRAIT (EMNLP '24) for reliable text generation. Designed synthetic data generation pipelines for mid-training and architected multimodal user & agent memory for Copilot Shopping (2 US patents). Fellow at Linear Capital (MLOps for foundation models). Research & teaching assistant at Stanford (NLP & scalability). ML Scientist intern at PayPal (RL & DeFi).
Stanford MS in Computational & Mathematical Engineering, minor in Philosophy. Previously at Erasmus University Rotterdam.
Featured Work
Projects
Interactive Guide
The Agentic Developer
An interactive exploration of Claude Code vs. Cursor — comparing philosophies, workflows, and hands-on agent building.
Explore →Community Guide
Claude Code Tips & Tricks
Curated tips from the community for getting the most out of Claude Code — from beginner to advanced techniques.
Read tips →Research
Entity-Aware Summarization
Entity-aware summarization for reliable sponsored search, ensuring AI-generated summaries align with actual webpage content.
View paper →