Founding Engineer

RunRL’s mission is to build the tools that make models intelligent, reliable, and aligned.

Responsibilities:

• Work directly with customers to integrate reinforcement learning into a multitude of workflows, from chemistry research labs to document processing teams at large enterprises
• Implement papers to experimentally validate new RL algorithms on customer problems
• Scale model training runs to hundreds of GPUs across many nodes
• Develop our self-serve platform, designing the frontend, integrations, and pushing the ease-of-use/capability frontier to democratize access to the best RL
• Run interesting experiments and write blog posts about the results
• Build high-quality internal tooling that speed up our team’s ability to move as fast as possible

You may be a good fit if you:

• Are excited about working at a fast-paced, rapidly growing company, and influencing its future trajectory
• Have a strong bias for action, and prioritize writing fast, high-quality code
• Have expertise in Python and Pytorch
• Possess excellent written communication skills and the ability to present findings
• Are passionate about developing the future of AI while solving problems that start out ambiguous and ill-defined

Strong candidates may also have:

• Experience working at developer-focused companies or on products targeting developers
• An understanding of the current landscape of reinforcement learning on language mdoels
• Experience deploying models on multi-node GPU clusters
• Familiarity with AI model outputs, as well as the ability to balance the use of AI coding assistants with the need for high-quality code
• Knowledge of AI alignment topics such as reward hacking, mechanistic interpretability, and capability evaluations

Logistics

We’re a small team working in-person in San Francisco. We value having a highly collaborative, optimistic team. Our primary goal is growing to a point which allows us to take much larger bets on AI-automated research and development. We all work from the RunRL headquarters in Hayes valley.