An open-source academic collaboration platform designed to streamline grant discovery and interdisciplinary team formation. Built to tackle funding fragmentation, it features AI-assisted grant matching, secure team workspaces, and a global research feed to help principal investigators and researchers advance science together.

An end-to-end real-time machine learning pipeline that transforms raw historical parquet data into a live event stream. It leverages Kafka and PySpark for data processing, applies an XGBoost model for fare and tip predictions, and serves insights through an interactive Streamlit dashboard.

A fully local RAG system that turns dense CS papers into concise, human-readable philosophical writing while keeping the technical core intact. Uses chunked document ingestion, HuggingFace embeddings stored in ChromaDB, and locally-hosted LLMs via Ollama to retrieve relevant passages and generate creative reinterpretations without any external API calls.

An end-to-end vision app that classifies clothing images in real time using a custom PyTorch model, a FastAPI backend for inference, and a React frontend for live camera capture. Trained on Fashion-MNIST with custom augmentation, the pipeline handles image preprocessing, model prediction, and result display with sub-second latency.

A retrieval-driven assistant that combines embeddings, reranking, and LLM reasoning to answer queries grounded in source documents. The system ingests documents into a vector store, retrieves the most relevant chunks using semantic search, applies a reranking step for precision, and feeds the result into a language model to produce accurate, citation-backed responses.

A marketplace platform for listing, searching, and transacting products with a focus on speed, clarity, and day-to-day usability. Features include advanced search with filters, user authentication, product CRUD, responsive grid layouts, and CI/CD via GitHub Actions for automated testing and deployment.

A RESTful API built with Spring Boot to power the Open Market Hub marketplace. Handles user authentication, product catalog management, order processing, and payment flows. Uses Hibernate ORM with MySQL for data persistence, Swagger for interactive API documentation, and follows layered architecture patterns for clean separation of concerns.