Founding ML Researcher

About the role

Base Compute is building the next generation of on‑device AI. As a Founding ML Researcher you will shape the research agenda, own experiments, and turn insights into real‑world performance improvements.

Key responsibilities

• Conduct inference research to discover new on‑device efficiency techniques such as speculative decoding, novel quantization, and other breakthrough methods.
• Develop model routing strategies that decide when to serve requests on‑device versus via frontier APIs.
• Design autonomous research pipelines that hypothesize, test, and evaluate ideas to accelerate the R&D loop.
• Create rigorous evaluations and benchmarks that reflect real‑world performance outside clean academic settings.

Required profile

• PhD in Machine Learning or equivalent industry research experience.
• Deep understanding of large‑language‑model architectures and AI inference principles.
• Proven track record of producing impactful results (papers, open‑source projects, or blog posts).
• Excellent communication skills for clear documentation and feedback.

Required skills

• Speculative decoding techniques.
• Quantization theory and practice.
• Model distillation methods.
• Reinforcement learning for inference optimization.
• GPU and accelerator programming (CUDA, ROCm, Metal, Triton).

What we offer

• Founding team equity and a strong base salary.
• Direct influence on technical direction and roadmap.
• Opportunity to solve genuinely hard, unsolved problems.
• Small, fast‑moving team with low bureaucracy.