Research Scientist, Reasoning
Make AI reasoning verifiable: methods that break problems into checkable claims, check them, and say honestly where they might be wrong.
- Research
- San Francisco (hybrid) or remote in the US
- Full-time
- $190k–$280k a year
What you will do
- Design and run experiments on decomposing problems into claims that can be checked one at a time.
- Develop methods for step-level verification with proof assistants, code execution and second models.
- Improve calibration, so Lemma's confidence matches how often it is right.
- Turn results into Lemma features, working closely with engineering.
- Write up what works and what does not, for publication and for our own record.
What we look for
- A track record of original research in machine learning, formal methods or a related field, shown in papers, code or shipped systems.
- Strong Python and experience running experiments with large language models.
- Care for rigorous evaluation, including negative results.
- Clear writing and the habit of making your reasoning easy to check.
Nice to have
- Experience with Lean, Coq or Isabelle.
- Work on chain-of-thought faithfulness, process supervision or calibration.
You do not need to meet every point to apply. We care about what you have built or proved more than titles.
Pay and benefits
The base salary range for this role is $190k–$280k a year. It is the range we expect in good faith to pay; where an offer falls depends on experience and skills. Offers also include equity.
- Compute budget for your own research questions
- Publish by default, with credit to the people who did the work
- Remote or hybrid, with team weeks in person
- Health insurance and generous parental leave
How to apply
Email contact@theoratix.ai with the role in the subject line, a link to your CV or LinkedIn, and something you have built or proved that you are proud of.
- Introductory call. A 30-minute conversation about your work and what you want to do next.
- Technical deep dive. Walk us through something you built or proved, and the decisions behind it.
- Work session. A paid, realistic problem, worked on together rather than set as homework.
- Conversation with the founder. Mission, values and the questions you have for us.
Theoratix Inc. is an equal opportunity employer. We welcome applicants of every background and do not discriminate on the basis of race, color, religion, sex, sexual orientation, gender identity, national origin, age, disability, veteran status or any other protected characteristic. We will consider qualified applicants with arrest and conviction records in accordance with the San Francisco Fair Chance Ordinance. If you need an accommodation during the hiring process, email us.