01GitHub live
Multi-agent reinforcement learning
MARL Soccer Self-Play
A lightweight 2D football environment for testing continuous-control agents, reward shaping, and self-play training strategies.
The lab
Things I'm building
and testing.
Smaller public projects where I explore an idea, test a method, or turn analysis into a usable tool.
Live GitHub metadata · refreshed twice daily02GitHub live
Statistical sampling
Sampling & Exploration Dashboard
A Streamlit workspace for exploring datasets and comparing simple random sampling with proportional stratified sampling.
03GitHub live
Applied data product
COSEM Dashboard
A Flask platform that combines public information, authentication, and private crop-analysis dashboards in one operational interface.