All Projects

A production-grade behavior-change application that combines behavioral science content delivery with an advanced AI-powered Retrieval-Augmented Generation (RAG) pipeline. The system transforms aspirations into systematic daily actions through personalized, science-backed recommendations delivered via a sophisticated ML backend and modern React frontend. Features include vector similarity search with FAISS, neural re-ranking, and keyphrase extraction.

A production-grade web application processing over 1 million ecosystem records to visualize Georges Bank & Gulf of Maine dynamics. Built for marine biologists and ecosystem scientists, it features interactive Leaflet maps with custom BGM coordinate transformation, advanced Recharts visualization suites, and a multi-dimensional filtering system. The platform handles large-scale time-series data through a highly optimized Flask/PostgreSQL backend with composite indexing and Redis caching, achieving sub-second query performance.

A comprehensive full-stack career platform revolutionizing the job search experience. It combines an applicant tracking system (ATS), AI-powered resume builder, and intelligent career coaching into a unified ecosystem. Features include drag-and-drop Kanban boards, real-time application tracking, automated resume parsing (PDF/DOCX) with ATS scoring, and gamified goal tracking to maintain user engagement.

A production-grade AI application addressing the reproducibility crisis in research. It reduces manual analysis time from hours to minutes using an 8-step automated pipeline. Features a multi-model LLM architecture (Claude, Gemini, Mistral) for analyzing papers against GitHub repositories, validating ground truth against DBLP/arXiv, and generating comprehensive reproducibility scores.

MetroShield AI is an intelligent real-time crisis response platform designed to simulate, detect, and visualize urban emergencies. Acting as a central command bridge, it ingests high-velocity data streams from simulated city sensors and calls, using Google Gemini AI to instantly analyze threats, assess severity, and generate rapid response strategies. The platform visualizes these incidents on an interactive, real-time 3D map, providing decision-makers with critical situational awareness.

A scalable big data pipeline designed to process unstructured job descriptions (JSON, TXT, PDF) from the City of Los Angeles. It enables structured querying and visualization by transforming raw data into a structured format using PySpark and AWS services. The system features a unified schema, automated text extraction using UDFs, and a robust data lake architecture on AWS S3.