Founding AI/ML Engineer

All rolesFounding AI/ML EngineerEngineering · Full-time · Hours overlapping San FranciscoApply for this role Own the agent loop and the evals that tell us whether a change made it better.ByteAsk is only as good as the decisions the agent makes between tool calls. You will own that loop: how it plans, which tool it reaches for, how it reads what comes back, and how we know whether a change made it better or worse.Most of the work is building evaluations from real C and C++ bugs and then moving the numbers on them. As one of the first engineers you will own the model side of the product, not a slice of it.What you’ll doBuilding and running evals from real C and C++ issuesImproving how the agent plans, picks tools, and reads their outputContext management across large codebases and many translation unitsModel selection and routing, and the cost versus quality tradeoffs behind themFine-tuning and data work, where it beats promptingYou may be a fit ifYou have shipped LLM or ML systems that real users depend onYou treat evals as part of the product, not an afterthoughtYou can read C or C++ well enough to judge a fix, or want to learn fastYou are comfortable owning an ambiguous problem with little guidanceYou would rather settle an argument with a measurement than an opinionApply for this role