Build agents that do useful work.
Design how agents reason, use tools, and work with people. Help us turn evolving AI capabilities into dependable software for complex workflows.
See open rolesWork on problems where correctness matters.
Building an agent means deciding how it gets context, chooses actions, handles failures, and knows when it needs help. Our work spans applied ML, system design, and product engineering, with close attention to how the whole workflow behaves in use.
Meaningful technical depth
Work on agent orchestration, context management, tool use, evaluation, and the infrastructure behind them.
A short path from research to product
Test ideas against real workflow constraints and turn what works into dependable product behavior.
Close collaboration
Engineers, designers, and domain experts work directly together on the same customer problem.
Real ownership
Own important decisions from problem definition through deployment and learning.
How we work.
Evidence over performance theater
We care whether a result is supported, reproducible, and useful under review.
Technical depth with practical judgment
We go deep when the problem demands it and keep the solution simple when it does not.
Direct, low-ego collaboration
We surface disagreement early, explain our reasoning, and improve the work together.
Respect for customer context
We protect sensitive information and never turn a private workflow into public marketing without permission.
Open roles
We are hiring software and machine learning engineers at Engineer and Senior levels, plus a senior go-to-market hire. Apply to the area that fits you and tell us what you have built.
Machine Learning Engineer
EngineeringSan Francisco · on-siteBuild and evaluate agents that reason over complex information, use tools, and complete multi-step tasks. Turn experiments into behavior we can measure and improve.
View role →Software Engineer
EngineeringSan Francisco · on-siteDesign the systems that let agents plan, call tools, and carry work through to completion. Build the runtime, integrations, and interfaces that make those workflows dependable.
View role →Senior GTM
Go-to-marketSan Francisco · on-siteHelp customers put AI agents to work on complex business workflows. Own discovery, product evaluations, and the commercial path to an agreed deployment.
View role →None of these quite you? We are always interested in exceptional people working where AI meets engineering practice.