Projects
Selected work across ML, GenAI, and deployable systems
A collection of projects focused on retrieval-augmented generation, OCR-based document intelligence, regression systems, forecasting, and deployment-oriented backend engineering.
CivicAI
A full-stack AI-powered municipal by-law assistant that combines RAG, pgvector semantic retrieval, Redis and Celery background processing, S3-based document storage, and FastAPI backend services to deliver structured, citation-backed answers through a production-style architecture.
CivicAI
A full-stack AI-powered municipal by-law assistant that combines RAG, pgvector semantic retrieval, Redis and Celery background processing, S3-based document storage, and FastAPI backend services to deliver structured, citation-backed answers through a production-style architecture.
InboxPilot AI
A full-stack AI-powered email assistant that integrates with Gmail and Google Calendar to classify job-related emails, track application status, automate interview scheduling, and generate an AI-driven career digest. Built with Next.js, FastAPI, PostgreSQL, and OpenAI, and deployed on Vercel and Render.
InboxPilot AI
A full-stack AI-powered email assistant that integrates with Gmail and Google Calendar to classify job-related emails, track application status, automate interview scheduling, and generate an AI-driven career digest. Built with Next.js, FastAPI, PostgreSQL, and OpenAI, and deployed on Vercel and Render.
Belleville By-Law Assistant
An OCR-powered RAG assistant for municipal by-law documents built with Tesseract, MiniLM embeddings, FAISS retrieval, FastAPI, and Streamlit/Gradio interfaces, designed to turn scanned PDFs into citation-grounded legal question answering workflows.
Belleville By-Law Assistant
An OCR-powered RAG assistant for municipal by-law documents built with Tesseract, MiniLM embeddings, FAISS retrieval, FastAPI, and Streamlit/Gradio interfaces, designed to turn scanned PDFs into citation-grounded legal question answering workflows.
Car Price Prediction
An end-to-end machine learning project covering exploratory analysis, feature engineering, model comparison, and deployment as a FastAPI microservice. Random Forest achieved the strongest performance, with R² of 0.908 and lower error than linear baselines.
Car Price Prediction
An end-to-end machine learning project covering exploratory analysis, feature engineering, model comparison, and deployment as a FastAPI microservice. Random Forest achieved the strongest performance, with R² of 0.908 and lower error than linear baselines.
Stock Price Prediction: ARIMA vs LSTM
A time-series forecasting project on Tesla stock data comparing ARIMA and LSTM models. LSTM significantly outperformed ARIMA, reducing RMSE from about 43.25 to 13.19 and better capturing volatile price movement.
Stock Price Prediction: ARIMA vs LSTM
A time-series forecasting project on Tesla stock data comparing ARIMA and LSTM models. LSTM significantly outperformed ARIMA, reducing RMSE from about 43.25 to 13.19 and better capturing volatile price movement.