Deterministic evaluation environment for AI code reviewers covering bugs, security (OWASP), and architecture via FastAPI + OpenEnv.

AI-powered system for low-exposure route optimization using AQI, simulation, and intelligent decision-making

AI research environment that simulates the end-to-end scientific discovery process, enabling agents to analyze papers, generate hypotheses, design experiments, and validate results collaboratively

Gymnasium RL environment for AI-powered customer support triage — classify, prioritize, assign, and respond to emails under SLA pressure. Built for the Meta PyTorch Hackathon under the OpenEnv spec.

An OpenEnv benchmark testing the ability of AI agents to act as Site Reliability Engineers (SREs) by diagnosing and filtering raw production failure logs.

High-fidelity Reinforcement Learning environment for smart grids. Features a custom DC Power Flow physics solver and real-world AT&C telemetry to train AI in power distribution and fault isolation.

RunbookOps: Deterministic OpenEnv environment for SaaS incident triage, runbook-driven resolution, and agent evaluation.

Adaptive Clinical Trial Design RL Environment

A reinforcement learning agent that learns to intelligently shape electricity demand, reducing peak loads and optimizing energy consumption in real-time.

OpenEnv Hackathon SF

A production-grade OpenEnv environment for benchmarking RL agents on real-world data cleaning and schema engineering tasks.

PyTorch Training Optimization RL Environment

Chess OpenEnv Multi-Agent System

Agentic Reinforcement Learning Loop to make Scientific Discoveries on Mars

CNN based PPO agent and LLM based GRPO agent to play SMB on openenv wrapper using Leirbag-gabrieL's gym-super-mario-bros fork

Multi-zone disaster relief AI env for Meta PyTorch OpenEnv Hackathon. 4-stage pipeline: PyTorch ZoneScorerNet -> Triage -> Planner -> Action Agent. False SOS detection, cascading failures, airlift precision.

Deterministic AI evaluation environment for customer support agents with explainable scoring and real-world task simulation.

🐛 Real-world GitHub issue triage environment for AI agent training — built on the OpenEnv spec with 3 difficulty-graded tasks, shaped rewards, and FastAPI server deployable to HuggingFace Spaces.

A realistic RL environment for training LLM agents on enterprise email triage—featuring multi-step decision making, ambiguity handling, tool usage, and deterministic evaluation.

OpenEnv code-debugging RL environment for the Meta × PyTorch Hackathon — 13 graded tasks, FastAPI, Docker, Hugging Face Spaces