hercules · San Francisco
Skills: React, TypeScript, HTML/CSS, SQL, AWS, Docker
What are we looking for? Hercules https://hercules.app is looking for a senior or staff level product manager to own evals for our AI coding agent What will you do? - Your 1 and only goal is to make Hercules' AI coding agent cheaper, faster, and more accurate (our pareto frontier) - You will own Hercules Bench, our offline eval set - You will own Hercules online evals - You will leverage our 500k+ users, 1M+ code repos, and millions of chat threads to guide your eval creation - You will anonymize data to protect our customer's privacy What skills do we expect you to have? We expect candidates to be incredibly strong across all of the following areas: - Software fundamentals: You are building an AI coding agent. We expect you to be technical and able to code. We expect you to have a decent understanding of system design, architecture, data modeling, and API design. In saying this, you will work with an engineering counterpart to work on complex implementation - Evals. You have explicitly created and/or managed technical evals before. You should be able to work across a variety of different eval methodologies and deeply understand optimal reward modeling, - Product sense: strong intuition for the goals customers are trying to achieve and novel solutions to help them achieve them - Design sense: Track record of building simple, delightful UX/UI that makes it easy for customers to achieve their goals - AI context and harness engineering: strong experience with the most up to date context and harness engineering best practices like summarization/compression, prompting, prompt caching, subagents, memory, context management, tools, async tools, deferred tools, skills, etc - Yes, we know this industry is evolving rapidly and the "best practices" change every week, but yes, we are still actively looking for someone who has wrestled with these challenges at scale and has strong opinions about what works and what doesn't - Our core stack: TypeScript, React, (some sort of agent…
Sign in to see the original posting
Browse more developer jobs on RoleBurst