Hey there! 👋 I'm a fresh grad of Computer Science graduate from The Chinese University of Hong Kong. I enjoy turning fuzzy ideas into things I can actually run, then checking whether they behave as expected. Lately, I've been applying that same mindset to large language models.
When I'm not coding, you can find me exploring the latest trends in large language models, experimenting with new agent workflows and eval setups, sketching out ideas for my next project, or simply enjoying a quiet afternoon in a bookstore.
Outside of tech I enjoy reading and learning. A physical bookstore is my happy place. I try to be a lifelong learner, including topics that sit next to CS, like how people pick up language.
A few things I have built
- LLM-leakage-eval: final year project. An eval bench that throws jailbreak prompts at open models and checks whether a synthetic privacy store leaks.
- HierarchicalRL-robot-navigation: summer research. Hierarchical DQN + TD3 for navigation in ROS + Gazebo, built on an existing TD3 baseline.
- Shop.com: course full stack shop in React / Node. I built the admin side.
- AlarmNest: course Android alarm app in Kotlin. Fun to make. Not the path I am on now.
Right now
- Working through the IBM RAG and Agentic AI Professional Certificate and the Google AI Professional Certificate on Coursera. Slow and steady. Cheer me on!
- Picking up Lua basics on Codédex so I can start scripting terminal apps and lightweight embedded systems.
- From September to December 2026 I will be in Hong Kong for TARA (Technical Alignment Research Accelerator), a 14 week in-person program on technical AI safety. I am really looking forward to Saturday pair programming, working through transformers, interpretability, RL, and model evals with a cohort, and wrapping up with a project presentation. See you on those Saturdays!
- Currently reading Social Robots: Boundaries, Potential, Challenges by Marco Nørskov.
Let's connect and build something awesome together! You can reach me at mayachen@link.cuhk.edu.hk.