Projects
A selection of systems and research prototypes spanning medical AI, retrieval-augmented LLMs, and computer vision.
Medical Imaging AI
Task-Driven IQA Evaluation for Lung CT
Duke collaboration evaluating 13 IQA metrics across NLST, VLST, and DLCS for detection, classification, and harmonization tasks.
LLM / RAG
Oncology Question Answering with RAG
FastAPI + SentenceTransformer + Qdrant pipeline for domain-specific retrieval and grounded LLM responses.
GitHub repository
LLM / AWS Bedrock
Intelligent PDF Query System
Built a scalable PDF Q&A workflow with AWS Bedrock, Titan embeddings, S3, EC2, and Streamlit.
GitHub repository
Computer Vision
Real-Time Object Detection
Implemented lightweight object detection using MobileNet-SSD and OpenCV for real-time inference.
GitHub repository