AI Engineer (Edtech, 50L)

This role is for one of our partner companies, (seed stage) raised $2.5M

About

we are building Bloom, a one-to-one voice and whiteboard maths tutor designed to make learning feel genuinely interactive and personal.

We're looking for an AI Engineer to own the entire AI layer of Bloom. This is an early-stage, high-ownership role where you'll work directly with the two founders and a small engineering team to take AI systems from architecture to production.

What You'll Work On

• AI Agent Harness: Build the orchestration layer that drives the tutor's reasoning, context, pacing, and pedagogical decisions.
• Evaluation Systems: Design and run evals to measure accuracy, safety, engagement, and teaching quality across thousands of interactions.
• Voice Pipeline: Own the STT/TTS stack and optimize for low-latency, natural voice conversations.
• Whiteboard Intelligence: Build the system that decides what the tutor draws, writes, and adapts on the shared canvas in real time.
• Personalization & Memory: Build systems that track each student's progress and adapt lessons based on what they know and where they're struggling.
• AI Architecture: Work directly with the founders to define and build the AI architecture for the next 2–3 months.

What we're looking for

• You've built and shipped production AI agents used by real users.
• You've designed the agent harness/orchestration layer, not just integrated an LLM API.
• You have experience building LLM evaluation systems and thinking deeply about how to measure quality.
• You're a strong production engineer, with a solid backend foundation and the ability to write reliable, maintainable code.
• You're comfortable operating in a 0→1 environment, making architecture decisions with incomplete information and shipping quickly.
• You actively follow the evolution of AI agents, voice systems, and LLM infrastructure and have strong opinions about what works and what doesn't.
• You're comfortable owning a problem end-to-end, from defining it to shipping and measuring the outcome.