Hassam HassamGani

Hi 👋, I'm Hassam Gani

🎯 Columbia sophomore focused on applied ML end-to-end design, training, evaluating, and deploying models across vision, text, audio, and tabular data.
🚀 Recent work includes a PyTorch DQN that learns Snake via self-play, an OpenCV-based MacOS FaceID that trains in seconds without GPUs, and an agentic Columbia RAG assistant.
🧠 Skills in Python, C++, Java, React, C, Bash, TypeScript, HTML5, CSS, PyTorch, Matplotlib, Scikit-Learn, Langchain, OpenCV, NumPy, Pandas.
🎓 B.S. in Computer Science, Columbia University in the City of New York, Expected Graduation May 2028.

Project	Description	Tech Stack
[FaceGuard] (github.com/HassamGani/FaceGuard)	Invented a lightweight facial recognition security software that quickly trains on owner’s face within 6 seconds without GPUs.	Python, OpenCV, Bash
[SmartSnake] (github.com/HassamGani/SmartSnake)	Built a reinforcement learning agent in PyTorch that plays Snake, with a two-layer network using 11 inputs, 256 hidden units, and 3 outputs.	Python, PyTorch, Pygame
[myHomer] (hassamgani.com/projects#myhomer)	Developed a campus-aware chatbot that uses Retrieval Augmented Generation over 470,944 chunks of embedded Columbia resources.	Python, Langchain, ChromaDB
[cpyAnalyzer] (github.com/wunused/python-ast)	Pickle-based Machine Learning models contain high cybersecurity risks — so I built this analyzer to mitigate them.	Python, AST

Contributed to generating policies that correctly loaded 79.8% of benign pickle-based ML models while rejecting 100% of malicious ones by developing cpyAnalyzer.

Credited in preliminary draft of forthcoming research paper on practical use and vulnerability testing of custom-made pickle-based models.

Black Muslim Student Association, Events Committee Member
- Preached, advocated, and ensured Black Muslims have as much of a say in our affairs as any other by organizing 7 major events.

⚡ "The best way to predict the future is to invent it." — Alan Kay