We are hiring an Engineering Manager to lead and grow the team that builds AiPrise's AI agents. This is a player-coach role: you will manage, hire and develop engineers, and you will also stay close enough to the code to design agent architectures, review prompts and tool definitions, and debug a failing eval yourself. You will own how our agents are built, measured and shipped: the skills and tools they call, how they remember context across a case, and the evals that tell us whether a change made them better or worse.
Responsibilities
Manage a team of 5-6 AI and backend engineers, including weekly 1:1s, clear goals, honest feedback and career growth plans; Hire the next wave of AI engineers and own the interview loop for agent-engineering roles; Run planning and delivery for the AI roadmap with product; Build a culture where engineers write evals before claiming a win and incidents lead to fixes, not blame; Design and review compliance agents, from single document checks to multi-step investigations that hand off to human reviewers; Build reusable skills, tools and memory for agents to call registries and data vendors and carry context across a case; Own the eval stack, including golden datasets built with compliance experts, regression tests in CI and production monitoring; Write and review code about 50% of the time, and choose models based on accuracy, cost and latency
Requirements
2+ years managing engineers directly, including hiring, performance reviews and at least one hard conversation handled well; A track record of growing engineers, such as a managed engineer being promoted or taking on bigger scope with your support; Experience running planning and delivery for a team in a fast-moving startup or scale-up; 6+ years as a software engineer with strong backend skills in Python or TypeScript; Built and shipped LLM agents to production, and can explain what broke and how you fixed it; Practical experience designing tool and function-calling interfaces, packaging reusable skills or prompt modules, implementing agent memory, and building evals that gate releases; Comfort with at least one major model API (Anthropic, OpenAI or similar) and the trade-offs between agent frameworks and plain code; A habit of measuring, with an eval you built, the metric it tracked and a decision it changed; Clear written communication, able to explain an agent failure to a compliance lead and to an engineer; Good judgment on when to build, when to buy and when to keep a human in the loop
Nice to have
Experience in fintech, regtech, fraud, identity verification or another regulated domain where AI decisions need an audit trail; Document AI experience: OCR, document classification or extraction across many languages and formats; Experience with LLM observability and tracing tools, and with cost and latency optimization at scale; Contributions to open-source agent, eval or retrieval projects, or public writing on the topic