Founder & Developer @ CellRepair AI
Building AI agents that make cellular health understandable for everyone.
AI debate agent independently validated at 96.5% win rate (201/208) using the RDI Foundation's AgentBeats debate evaluation framework.
Outperforms GPT-4, Claude 3.5 Sonnet, and Gemini Pro in structured patient education debates.
| 96.5% Win Rate | 201/208 Debates Won | #1 AgentBeats Ranking |
Patient-friendly cellular health education powered by MedGemma 1.5 4B. Submitted to the $100K MedGemma Impact Challenge on Kaggle.
| 4.68 / 5.0 LLM-as-Judge Quality | 4B params Edge Deployment | 8 GB VRAM Runs Locally |
🔒 Runs on edge hardware — no cloud, no patient data leaves the device.
A network of autonomous AI agents for health education with integrations across the AI ecosystem:
🔗 LangChain · 🤖 Claude MCP · 🤗 Hugging Face Space · ⚡ Grok Integration
| Project | Metric | Result |
|---|---|---|
| AgentX Purple | AgentBeats Validation | 96.5% (201/208) |
| Health Educator | LLM-as-Judge Quality Score | 4.68 / 5.0 |
| Health Educator | Edge Deployment (4B params) | 8 GB VRAM |
🟢 cellrepair.ai · 🎬 Demo Video · 🤗 Hugging Face Space · ⚙️ AgentBeats Framework