ML & Neural Networks

Projects applying machine learning, computer vision, and data analysis — from training CNN agents to play RTS games, to building recommendation systems from large-scale music datasets.

CNN StarCraft Agent
▶
CNN · Python
CNN Computer Vision Game AI

CNN Agent for StarCraft II (Minimap-based)

A Convolutional Neural Network trained to play a StarCraft-like RTS by processing minimap frames as visual input. The model learns to map spatial patterns — unit positions, movement, map control — directly into game actions and commands, enabling real-time strategic decision-making from pixel data. The project covers data collection from gameplay sessions, minimap image preprocessing, model training and evaluation, and an inference loop integrated into the live game environment.

Python PyTorch CNN Computer Vision StarCraft II API NumPy Matplotlib
Spotify Visualization
▶
Python · Streamlit
Data Analysis Visualization Recommendation Systems

Music Data Exploration & Recommendation System using Spotify Data

An end-to-end data science project exploring large-scale Spotify music data spanning 1921–2020 through interactive visualizations and trend analysis. The project uncovers how musical features — tempo, energy, valence, acousticness — have evolved over time, and uses these audio characteristics to build a content-based recommendation system that suggests tracks based on sonic similarity to a given song.

Python Pandas Scikit-learn Matplotlib Seaborn Streamlit Content-Based Filtering