FuelNetEnv

An OpenEnv-compliant RL and LLM-agent environment (FastAPI) that simulates global fuel distribution during real geopolitical crises, with five graded scenarios up to a full Strait of Hormuz closure. Its hybrid baseline pairs a deterministic logistics engine with Llama-3.1-8B, scored 70% deterministically and 30% by an LLM judge. Built at the Meta PyTorch OpenEnv Hackathon as team Dhokla Deep Learners.