← All roles

ML Research Scientist

Theorem· Engineering, Machine learning
Apply on company site
San Francisco, CA$300k - $500kFull-timePosted 11mo ago

About the role

Theorem is building AI-powered program verification tools to improve software correctness, understandability, and security. The ML Research Scientist role focuses on post-training model capabilities for program analysis and formal reasoning, enabling scalable correctness feedback and compiler-level reasoning to help users find bugs and generate design documentation.

Responsibilities

  • Work on post-training models to improve program analysis and formal reasoning capabilities
  • Develop interfaces to navigate complex implementations of simple specifications for bug finding and documentation generation
  • Contribute to compiler-level reasoning methods to support non-functional requirements and system reliability
  • Tackle last-mile problems and work on projects that are large, fast, or complex compared to state-of-the-art
  • Collaborate with founders and cross-functional team to advance product goals

Qualifications

  • 3+ years of experience in relevant field
  • Experience in machine learning applied to program analysis or formal methods (preferred)
  • Strong analytical and problem-solving skills
  • Ability to explain complex concepts clearly and provide satisfying explanations of how and why things work
  • Willingness to tackle large-scale, high-impact projects
  • Experience in research or industry roles related to ML for software correctness (preferred)

Skills

Required:Machine learning for code analysisCommunication and collaboration
Nice to have:Formal reasoning / formal methods

Benefits

  • Competitive compensation with equity
Last updated Aug 26, 2026. Interested? Apply on the company site.
Theorem logo

About Theorem

Program verification so even your systems engineers can vibecode

Theorem is training models that make program verification 10,000 times faster. Using verification as a feedback loop, developers have found zero-days in GPU accelerated code and cryptography implementations, and sped up code migration in legacy systems. If you have complicated code that needs to be correct and secure, sign up for our beta!

machine-learning

More open roles at Theorem