Credit risk, fraud, seizure detection. I take a problem from raw data to a model you can question, then put it somewhere people can use it. BS Computer Science graduate (2026); IBM Machine Learning Professional Certificate.
Portfolio · LinkedIn · muhammadhasaan139@gmail.com
Each one runs in the browser, and each is built as its own app.
Also on the site: Nimbus (weather), Punchlist (tasks), Tally (calculator).
- webotMCP — MCP server for the Webots robot simulator: control robots, edit scenes, read sensors and cameras from an AI assistant.
- kpi-nexus-final — multi-tenant KPI SaaS. NestJS, Next.js, Prisma, TimescaleDB, pgvector, BullMQ and a Python ML sidecar.
- Seizure detection — pose estimation that watches for seizure-like movement and alerts carers. OpenCV, MediaPipe, TensorFlow.
- Malware detection — parallel scanning and classification of malware signatures with Python multiprocessing.
- PDF_Scrapper — receipt PDFs to clean CSV.
Modelling Python, SQL, PySpark, DuckDB, XGBoost, scikit-learn, PyTorch, TensorFlow, SHAP
Serving FastAPI, Django, Docker, n8n
Web TypeScript, React, Next.js, three.js