Curriculum Engineer
We're working with an early-stage AI startup building the training infrastructure that frontier AI labs use to train their agents. Still in stealth
They build pixel-perfect clones of professional software (think Gmail, Slack, Salesforce, QuickBooks - that kind of everyday tool) and turn them into training environments. On top of these clones, they build curricula of tasks with deterministic graders, which they sell to some of the biggest AI labs in the world to train their agents. $10M in revenue in the first 100 days, $50M raised, ~100 people across Paris and London.
Where they're going: Targeting $100M ARR by end of 2026, with a plan to hire ~400 people over the next 12-18 months. This is still very early - the team, the process, and the org are being built in real time, and early hires shape a lot of that.
Responsibilities
- Design and calibrate training tasks on cloned software environments
- Write deterministic graders - no LLM-as-judge, binary pass/fail
- Calibrate task difficulty precisely (targeting a specific pass rate window on frontier models, not too easy, not impossible)
- Iterate fast, QA manually, and dig into failure modes others haven't found yet
Required Skills
- Solid software engineering fundamentals (professional SWE experience or equivalent)
- Some exposure to LLMs / agentic systems (using LLM APIs counts)
- A hacker/experimenter mindset - comfortable with rapid iteration and manual QA
- Strong intellectual curiosity - you'll be learning a new domain on every project
- Work at the center of frontier AI, on infrastructure the biggest labs in the world depend on
- Explosive traction: $10M revenue in 100 days, $50M raised, $380M valuation in under a year of existence
- Fast progression: €20-40k raises already observed internally for engineers who level up
- Multiple career tracks after the initial ramp-up: senior Curriculum Engineer, AI Engineer, Software Engineer, or leadership/management
- Direct, fast impact on a small, highly agile team
Practical info:
- Based in Paris (3rd arrondissement) or London
- On-site full-time for the first month, then 3 days remote/week