Build AI to co-invent the future

Our mission is to empower people to invent solutions for the world’s hardest problems, working together with scalable and reliable AI agents. For example, our agents can efficiently reason about and verify mission-critical code, including https://asari.ai/blog/space-grade-software.

Our team has published award-winning AI research and is backed by top-tier investors including Eric Schmidt, Caltech, Jeff Dean, and JP Millon.

We move fast, think from first principles, and build with purpose. We believe that great results come when people take ownership, grow together, and share both the challenges and the wins.

What you'll do

Bring deep kernel expertise to our AI agents that optimize high-performance, mission-critical computing systems. You'll shape the tools and feedback loops those agents use to improve performance across real workloads. You’ll work at the intersection of kernel engineering and applied AI to scale up AI agents that design and build complex systems.

What you have

• Strong CUDA C experience, with hands-on work implementing or optimizing kernels for ML or other GPU-heavy workloads
• Fluency in Python and C/C++, and familiarity with the ML kernel ecosystem (PyTorch, Triton, JAX, as well as other frameworks such as CUTLASS, cuBLAS)
• Deep knowledge of NVIDIA GPU architecture, performance tooling, and profiling
• Exposure to low-level primitives and tools like PTX and XLA
• Excellent analytical thinking from first principles and communication skills

Nice to have

• Ability to work effectively with AI coding agents like Claude Code
• Experience with TPUs, AMD GPUs, edge AI, or other accelerator ecosystems
• Familiarity with distributed systems

Compensation, benefits, and perks

We offer competitive compensation + stock options, full health coverage, and a 401(k) match, plus additional wellness perks, flexible hours, and unlimited PTO. We prioritize growth and connection through daily meals in office, learning budgets, and regular team socials.