I built a RAG-Postmortem Intelligence System that indexes incident logs and past postmortems, then answers root-cause questions with cited evidence. I combined keyword and semantic search so answers trace back to real data.
For the system, I designed a two-service architecture with Django for authentication and multi-tenancy and FastAPI for the AI pipeline. I containerized it with Docker Compose and replaced paid embedding APIs with local embeddings.
For my Brain Tumour Detection System, I trained an EfficientNet-B3 model on 7,023 MRI scans, reaching 95.65% accuracy and a 99.76% macro ROC-AUC. I paired predictions with Grad-CAM and Grad-CAM++ explanations for clinicians.
I also tuned a Random Forest model for a heart disease prediction web app, achieving 98.53% accuracy on 500+ clinical records, and deployed an interactive Streamlit dashboard. As Technical Head of FOCUS, I lead “Blind Coding 2026” and organize technical events.

