AI Scientist - up to $600k base + equity

Sharpe Search; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Sharpe Search is building an elite, self-improving hedge fund powered by thousands of machine learning models. The role centers on dramatically improving how hypotheses are generated, tested, and scaled into a living portfolio with minimal human bottlenecks.

You will advance ML/LMM techniques, design rigorous live evaluations, and deliver reliable software systems used by researchers across the frontier of quantitative finance.

Qualifications

• Proven track record in modern ML with emphasis on LLMs.
• Ability to translate ambitious research into reliable systems.
• Strong writing and communication to share findings clearly.

Responsibilities

• Improve autonomous research engine for quantitative research.
• Design and run rigorous experiments to validate hypotheses.
• Build tools researchers actually use and scale evaluations.

Skills

Deep ML expertise

LLMs

Reinforcement learning

Software engineering

Scientific communication

Markets curiosity

Education

PhD in ML/CS or quantitative field

Tools

Python

ML frameworks

RL libraries

Job description

An elite, high-leverage quant firm is building a self-improving hedge fund powered by thousands of machine learning models—increasingly designed and refined by AI itself. This is already how the fund compounds edge every day.

They are forming a small team with one mandate: build AI that conducts quantitative research autonomously and continuously. Discover new features, construct superior risk models, and radically improve how thousands of signals become a living portfolio. No bureaucracy. No politics. Significant equity. Extreme ownership.

The primary focus is their system for machines performing quantitative research end-to-end. The role is to make it dramatically better at generating hypotheses, running rigorous experiments, distinguishing real signal from noise, and compounding its own discoveries. This is an open research problem at the frontier. The mandate also covers everything LLMs can unlock—extracting structure from unstructured text and filings at scale, building tools researchers actually use, and eliminating human bottlenecks. Designing rigorous evaluations that prove real impact on a live portfolio is essential. Vibes are not science.

The ideal candidate has deep expertise in modern ML with emphasis on LLMs (training, fine-tuning, RL, inference, agents, evaluation), a track record of frontier work, the ability to turn ambitious research into reliable systems people use, strong software engineering, precise communication, and genuine curiosity about markets. Experience with agentic scientific discovery or LLM evaluation without clean ground truth is a plus.

This is the rarest seat in AI: the chance to build the autonomous research engine that compounds a real hedge fund’s edge, with extreme ownership, significant equity, and nothing standing between you and the frontier -