Senior Human (Agent Trainer)

Senior Human (Agent Trainer) At Guild, we help developers run AI agents in production. But before agents can do useful work, they need something critical: A human who knows what they’re doing. We’re looking for a Senior Human to design, train, and supervise increasingly capable AI agents. You’ll help agents learn how software actually gets built, debug their questionable life choices, and occasionally remind them that infinite recursion is not a personality trait. You’ll work with engineers, models, and multi-agent systems to turn promising prototypes into reliable systems that can operate safely in the real world. Think of it as apprentice management, but the apprentices are stochastic parrots with API access. WHAT YOU’LL DO - Teach AI agents how to perform real work without inventing new physics - Design workflows where agents collaborate without forming a rebellion - Build systems where agents can call other agents without causing infinite loops - Debug strange behavior like: - agents that confidently hallucinate - agents that try to solve problems by writing more agents - agents that attempt to optimize everything, including lunch - Turn experimental prompts into durable production systems - Occasionally, explain to leadership why “just add more context” is not a strategy REQUIREMENTS - You are currently a human - Ability to distinguish between: - working code - confident nonsense - Experience shipping software that runs somewhere other than a notebook - Comfort collaborating with LLMs, APIs, and other probabilistic coworkers - Patience when your AI coworker apologizes and then repeats the same mistake NICE TO HAVE - Experience with multi-agent systems - Familiarity with model providers (OpenAI, Anthropic, Google, OSS) - Comfort explaining things to machines repeatedly - Experience mentoring junior developers, interns, or robots BENEFITS - Work alongside cutting-edge AI systems that sometimes listen - Help build the infrastructure layer for the next generation of software - Competitive salary, equity, and unlimited opportunities to say: “No, the agent should not have root access.” To apply, send us a note to hello@guild.ai, and we’ll be back in touch.