Archive
All Projects
A complete catalog of data engineering pipelines, AI systems, and web applications.
FIFA Soccer DS
Multi-model CV pipeline for soccer video analysis with YOLOv8n detection, ByteTrack tracking, and GraphSAGE interaction classification. Full MLOps with DVC, MLflow, and FastAPI.
Soccer Vision Research
Research framework for multi-model soccer video analysis combining RF-DETR detection, SAM2 segmentation, and SigLIP zero-shot identification.
WebMCP Portfolio Integration
Making this portfolio AI-agent queryable using the W3C WebMCP standard — one of the first implementations.
Nobel Data Intelligence
Physics-informed deep learning for protein stability and enzyme kinetics. Tri-modal architecture combining ProtT5, GATv2+VDOS, and ChemBERTa with gated attention fusion.
Website Watcher
Production web archiver with automated crawling, SHA-256 change detection, FTS5 full-text search, and cryptographic verification. 26 Python modules with Docker orchestration.
CAG Deep Research
Production multi-agent research automation system built in 10 days, orchestrating 5+ specialized AI agents with LangGraph to autonomously synthesize 20-page research reports from raw queries.
ContextBox
CLI-first personal knowledge assistant. Captures screenshots with OCR, embeds text with sentence-transformers, and enables semantic search + AI Q&A from the terminal.
Kayak - Travel Metasearch
3-tier distributed system simulating Kayak's travel booking with microservices, caching, and AI recommendations.
Airbnb - Distributed Booking
Cloud-native Airbnb clone with microservices, Kafka messaging, and AI-powered travel planning.
Rubik's Cube Timer & Visualizer
Keyboard-first speedcubing timer with Three.js 3D visualization, WCA-compliant scrambles for 2x2–7x7, session management, and full PWA support.
CPU Scheduling Visualizer
Interactive visualization of 9 CPU scheduling algorithms with Gantt charts and comparative analysis. Applied to vaccine scheduling in a published IEEE paper.
Diabetes Prediction (Stacking)
Stacking ensemble classifier for early diabetes prediction using 6 base models. Published at AIMV 2021 on IEEE Xplore with 10 citations.
Stock Data Platform
Unified market data warehouse — 10 tickers, 25 years of history, 4 data sources. Kafka streaming, Airflow orchestration, TimescaleDB star schema, Dockerized.
Wikipedia Analysis
URL → scrape → tokenize → top frequent words with visuals.