How to apply for ML Research Engineer

About Theorem

Theorem is an AI company building systems for formal program verification to improve software reliability. It operates as a programming languages and AI lab, with a team from cryptography, infrastructure, security research, and top-tier companies. The work targets correctness feedback for AI software engineering at arbitrary complexity.

About the role

As an ML research engineer, you will conduct research and develop products that enable AI models to perform software engineering tasks beyond current human capabilities. You will work on problems orders of magnitude larger, faster, or harder than state-of-the-art, solving last-mile technical challenges. The role is remote and full-time, with multiple related positions available.

A typical day

A typical day might involve researching new methods to apply formal verification to AI-generated code, prototyping tools, and collaborating with a remote team on technical challenges. You could also spend time debugging complex systems or writing papers. Specific routines are not provided; ask the team about daily workflows during interviews.

Who Theorem is looking for

• Background in machine learning research, especially applied to program synthesis, formal methods, or code generation.
• Experience with programming languages, compilers, or formal verification techniques.
• Proven ability to solve difficult technical problems at scale, such as those involving large codebases or complex systems.
• Comfort working in a remote, research-driven environment with engineers from cryptography, infrastructure, and security backgrounds.

Tips for this application

• Highlight any work you have done on formal verification, program analysis, or applying ML to software engineering tasks.
• Show evidence of tackling problems that are orders of magnitude larger or harder than typical, such as scaling systems or optimizing performance.
• Mention specific programming languages and tools you have used for verification or AI research.
• Explain why Theorem's focus on formal program verification for AI software engineering interests you.
• If applying for the research fellowship, clarify your availability for a 3–6 month period.

What to cover in your cover letter

['Your experience with formal methods or program verification and how it applies to AI-driven software engineering.', 'Concrete examples of solving last-mile technical challenges in ML or software engineering.', 'Your ability to conduct research and develop products that scale correctness feedback.', 'Why you want to join a lab focused on programming languages and AI, not just a general AI company.']

Research before applying

• Read Theorem's published research or product announcements on formal program verification and AI software engineering.
• Understand the basics of formal verification and its role in improving software reliability.
• Look into the backgrounds of the team members to see their work in cryptography, infrastructure, and security.
• Find out what 'correctness feedback at arbitrary complexity' means in practice by checking their blog or papers.

Likely interview topics

Based on the job description, expect questions about:

• How to integrate formal verification with machine learning models for code generation.
• Approaches to scaling correctness feedback for large, complex software systems.
• Your experience with programming languages, type systems, or static analysis.
• A technical challenge you solved that was orders of magnitude beyond state-of-the-art.
• How you would collaborate with engineers from cryptography, infrastructure, and security backgrounds.

Common mistakes to avoid

• Submitting a generic ML resume without highlighting formal methods or program verification experience.
• Focusing only on model training and ignoring the programming languages and verification aspects.
• Not explaining why you want to work on last-mile technical challenges rather than general AI research.