Obi-Wan Kenobi: “That’s no moon. It’s a space station.”
Ever wished you could take everything you learned at a frontier lab and point it at one of the hardest, most fascinating problems in science, with a blank whiteboard and a serious budget? Here’s your shot.
A well-funded, stealth-mode AI startup, founded out of several of the world’s top research universities and institutes, is hiring a Senior AI Leader to own its entire AI research agenda. The company sits at the intersection of frontier AI and fundamental science, and this hire decides how frontier-grade models and agents get built, trained, and scaled. From scratch. Your call.
The split is roughly 60% hands-on research and engineering, 40% leadership and strategy. Translation: you still get to train models and argue about architectures, but you also get to build the team and the culture around you. If you’ve shipped AI-for-Science work at a frontier lab and you’re itching to run your own show with real ownership and meaningful equity, keep reading.
What You’ll Do
- Set and drive the AI research roadmap across foundation models, reinforcement learning, and agentic systems
- Stay deep in the code: architect models, design training schemes, and prototype new agent behaviours alongside brilliant domain scientists and engineers
- Build the training infrastructure and evaluation frameworks that let AI systems generate, test, and validate results at scale
- Recruit and lead a world-class team of AI researchers and engineers, and set a high bar they’ll be proud to clear
- Shape how the org runs: project selection, experimentation, code review, cross-team collaboration
- Be the voice on AI internally, guiding technical trade-offs and long-term architecture decisions
- Partner with the founders to align AI strategy with company milestones and big scientific ambitions
You Might Be Our Person If…
- You’re at (or recently left) a frontier AI lab like OpenAI, Anthropic, or Google DeepMind, ideally with shipped AI-for-Science work. Other frontier labs considered case by case
- You know modern ML cold: foundation-model pretraining, RL (PPO/SAC-style methods, self-play, or similar), large-scale optimisation, and distributed multi-GPU/multi-node training (CUDA, NCCL, the works)
- You’ve managed and grown small, high-performing research or engineering teams, and enjoyed it
- You can take a wildly ambitious goal and break it into projects a team can actually execute
- You have a strong background in physics, maths, or the hard sciences. Pivoted from hard science into AI recently? Even better, we love a good origin story
- You’re happy in the trenches with domain scientists, debating benchmarks, evaluation harnesses, and what “valid” really means
Bonus points: scientific ML (physics-informed neural networks, neural PDE solvers, differentiable simulation, surrogate modelling), experience wiring models into large-scale simulations or complex simulators, or familiarity with export-control and safety considerations around advanced AI.
The Practical Bits
- Location: Boston, hybrid in-person. East Coast (e.g., NYC) works if you’ll be in the Boston office at least monthly; West Coast works if you’re up for relocating. Relocation support on the table
- Compensation: ~$400k base plus 3–5% equity at the seed stage, with competitive benefits. Yes, that equity number is real
- Work authorisation: US citizenship or green card strongly preferred; sponsorship considered case by case for exceptional candidates
- Note: This role is subject to US national-security and export-control requirements, which may limit eligibility for some candidates
