About the role
The AI Engineer will build the agentic systems at the core of the product: systems that understand each learner, plan a path with them toward skills worth having, and work with them step by step until they get there.
This is not a wrap-an-API role. The hard problems are the ones frontier models don't solve on their own: maintaining an accurate picture of a learner over weeks and months, deciding what to teach next and when to hold back, keeping long-running conversations useful rather than merely pleasant, and verifying that generated teaching is correct before a learner ever sees it. You'll own systems end to end - design, implementation, evaluation, and iteration against real learner data.
What you will do
- Design and build the agentic core: multi-step tutoring loops, tool use, memory, and planning over long-horizon learner relationships
- Build the learner model - the evolving, evidence-backed representation of what each learner knows, wants, and responds to - and the systems that read and write it
- Build evaluation harnesses for conversational quality and teaching quality, and use them to drive iteration; define what "this session taught something" means operationally and measure it
- Design guardrails and verification layers so generated content and tutor claims meet a bar a trusted brand requires
- Work daily with the founding team, including Andrew, on the hardest product questions: what should an AI tutor do, and how do we know it's working?
What you bring
- AI-native: you default to AI-assisted coding and building agentic automations in everything you do, you have an appetite for and record of experimenting with the newest AI engineering practices
- 3+ years as a software engineer, with substantial hands-on experience building with LLM APIs (Claude, OpenAI, or similar): agentic workflows, tool use, structured output, long-context and memory patterns
- Experience shipping and operating LLM systems in production, including evaluating them - you have opinions about evals because you've built them
- Strong Python and/or TypeScript/Node engineering skills; comfort owning services end to end
- Ability to turn a fuzzy product question ("is the tutor actually helping?") into a measurable system, and ship without heavy oversight
Nice to haves
- Experience with conversational AI products, tutoring systems, or long-running assistant relationships
- Background in recommendation, personalization, or user-modeling systems
- Familiarity with the education or learning-science landscape
- Experience with voice interfaces or real-time interaction
What success looks like
In your first 30 days, you will have shipped a measurable improvement to the core tutoring loop and stood up an evaluation that tells us whether it worked.
In your first 6 months, the agentic core - learner model, planning, verification - will be a durable system the whole product builds on, with quality metrics the team trusts and a cadence of improvement driven by real learner data.
