Cambridge
Hackathon

Apply Reinforcement Learning to Gaming.
Limited Spots Available This November.
Don't Miss Out.

Launch: 24th November 2025

Development (AI vs Computer): 24 Nov - 31 Dec 2025 (Online)

Finals (AI vs AI): Jan 2026 TBC (in person)

Finals Venue: Constance Tipper Theatre, Cambridge

Prize Pool: $3000 - Sponsored by 0g Labs, 2nd tournament in January 2026

* All phases online except the in-person finals showcase at Cambridge

📝

Step 1: Register

Fill out the registration form

📱

Step 2: Join WhatsApp

Connect with participants & get updates

Complete both steps to fully participate in the competition

What is DIAMBRA?

DIAMBRA is a reinforcement learning platform that transforms classic fighting games like Street Fighter, Tekken, and Mortal Kombat into standardized training environments for AI agents. It enables researchers and developers to build, train, and compete with AI agents using familiar RL frameworks, exploring competitive multi-agent systems, self-play, human-in-the-loop training, and imitation learning. The platform features a global leaderboard, live AI-vs-AI battles, and supports cutting-edge research in embodied AI and game-playing agents.

Competition Sponsors

Competition Details

🎮

What You'll Build

Design and train RL agents in Street Fighter III using the DIAMBRA platform. Your agents will compete in:

AI-vs-COM benchmarking
AI-vs-AI competitive matches
Live finals showcase

                            Special Prize: Best use of Mistral AI's LLM capabilities integrated with DIAMBRA's LLM module wins API credits!
                        

🧠

What You'll Learn

Gain hands-on experience in:

Reinforcement learning fundamentals
Multi-agent competitive systems
Model evaluation & deployment
Self-play & imitation learning
LLM integration for game AI

🏆

Who Can Join

Open to all students worldwide:

Undergraduate, postgraduate, PhD
Solo or teams (max 3 members)
Python familiarity recommended
No prior RL experience required

Top 10 agents advance to the in-person finals at Cambridge!

CambridgeHackathon