Portfolio

AI system that clusters and captions images using CLIP embeddings and FAISS search, with face recognition/clustering for a personal timeline. Drag-and-drop Streamlit UI supports grouping by objects or people.

AI-powered archaeology project from Hacklytics. Trained on 100k+ synthetic and 5k real fossil images; achieved ~99% accuracy for fossil identification/valuation.

Research paper on arXiv: Python-based approach to automate Knights-and-Knaves reasoning. Earned 3rd prize at the InnoSphere International Research & Tech Conference.

Custom-built personal portfolio with advanced text scrambling animations, fluid UI components, and responsive design. Features smooth transitions and professional navigation.