At Medecro.ai, I built and deployed a production tooth segmentation and numbering system for OPG X-rays, reaching approximately 97% numbering accuracy. Clinical users actively use it.
I also developed production detection models for RVG X-rays, identifying dental pathologies and structures across multiple classes. These pipelines are integrated into Medecro.ai’s imaging system to support automated clinical insights.
On my Surgical Agent project, I developed a laparoscopic surgery analysis pipeline that combines anatomical segmentation with surgical action recognition. It provides visual grounding for multimodal clinical reasoning and Critical View of Safety assessment.
I’ve also worked on biomedical question answering, road-scene analysis, and chest X-ray classification projects. My work spans computer vision and generative AI, from model fine-tuning to deployment with tools including PyTorch, ONNX, and FastAPI.
