AI/ML Engineer – Existing Working AI Training Code Required (RL, LLM, AlphaZero, MuZero, GRPO)

We’re looking for an experienced AI/ML engineer for a potential long-term hourly contract, starting with a small paid project. We’re interested in both LLM-based and non-LLM-based training projects. Examples include AlphaZero, MuZero, PPO, DQN, SAC, self-play RL, chess, Atari, GRPO, RLHF, DPO, Unsloth, TRL, OpenRLHF, ART, or any other proven AI training implementation. What we're looking for: - You must already have working and tested AI training code. We are not looking for something developed from scratch. - The code should demonstrate measurable improvement in model performance through training. - We need the full source code, setup instructions, and evaluation results. - We should be able to run the training ourselves and continue improving the model. When applying, please tell us what you already have. If you have a demo video showing the actual training process and results, include the link directly in your proposal. GitHub links, training logs, or other evidence would also be helpful. We want to move quickly, so we're particularly interested in candidates who can demonstrate their existing implementation immediately. Long-term opportunity: Our main goal is to find someone we can work with on AI/ML research and development over the long term. However, before offering a long-term hourly contract, we need to see that you can deliver a genuinely working training system. This initial project is paid on a fixed-price basis. We are not asking anyone to work for free. Successful delivery and demonstration of this project is a requirement for being considered for our long-term hourly contract. Please apply only if you already have relevant working code. Generic, AI-generated proposals that don't provide specific details or evidence of previous work will be ignored. We're looking to get started immediately and complete this initial project as quickly as possible.