I'm an AI Engineer who ships full-stack, AI-powered products end to end: RAG pipelines, LLM integration, and tool-calling agents that reason over real tools instead of guessing. My background in medical imaging shapes how I build. Rigorous, clinically-minded, and obsessed with turning messy real-world data into something trustworthy.
- 🤖 Building agents that verify their own answers. Every quote and citation is mechanically checked against tool output before it reaches the user
- 📏 Evals over vibes: 43-case agent benchmark at 97.7% tool selection and 96% groundedness, with tracing, CI and 168 tests behind it
- 🔬 MSc Medical Imaging Science, University of Manchester (Merit)
- ⚙️ BEng Biomedical Engineering, Shantou University (Top 30%)
- 🌐 Portfolio: joychen-ai.vercel.app
| 🩻 RadReport Agent Tool-calling agent for chest X-ray QA. It picks its own tools (classification, segmentation, CTR measurement, retrieval), then mechanically verifies every quote and citation against tool output before answering. 97.7% tool selection, 96% groundedness over 43 cases. Live demo →    |
🧠 Medical Image Analysis Platform Full-stack CT/MRI segmentation demo: React front end, ASP.NET Core API, FastAPI/MONAI inference service.    |
| 🍲 PantryChef Tell PantryChef what's in your kitchen and get matching recipes and cooking videos, a Nutri-Score-style health grade, and RAG-powered Q&A grounded in a meal-prep knowledge base. Live demo →    |
🩸 ICH CT Segmentation Automatic intracranial haemorrhage segmentation from non-contrast CT using U-Net-based workflows.  |
