Lead LLM Engineer
Help AI understand how work should be done. Lead the experiments, evaluations and engineering that make generated workflows more useful and dependable.
What you'll take on.
- Improve how we extract workflow structure from interviews, documents and examples, including missing information and conflicting instructions.
- Build evaluations for generated workflows: completeness, factual grounding, exception handling and whether the work meets the agreed standard.
- Compare prompting, retrieval, structured generation and fine-tuning against explicit baselines, including quality, latency and cost.
- Design datasets and experiments for specialized models, with clear provenance, appropriate permissions and separation of training and evaluation material.
- Turn promising results into maintainable product capabilities, document failure modes and work with implementation teams to test improvements.
Show us how you think.
Bring examples that show how you learn from model behavior and turn that learning into better software.
- You have built LLM applications and can trace an output failure back to the data, context, instructions, model or surrounding code.
- You have run model training or fine-tuning experiments and can explain the baseline, data choices and measured result.
- You write strong Python and can build reproducible evaluation and data pipelines.
- You can distinguish a persuasive demonstration from a repeatable improvement, and make that distinction clear to the team.
What early progress looks like.
A useful evaluation set for workflow building, a reproducible baseline and a small set of experiments that show which changes improve results. That evidence should guide both the next product release and the direction of model research.
The company you're helping build.
Fully Powered AI is an AI native services startup based in Los Angeles. We build our own technology to help companies capture the way their best people work and establish capabilities they don't yet have. Fully Powered Studio and reusable workflow foundations turn that knowledge into AI workflows teams can use, check and improve.
Our next ambition is to develop models specialized for building business workflows. You'll lead that research direction, including fine-tuning experiments, while improving what we can deliver with existing models today. The question is practical: what helps a model capture the judgment in a job and turn it into a capability someone can use?
Help build what comes after the prompt.
Send a resume or profile and one example of an LLM system or model experiment you worked on. Tell us what you changed, how you evaluated it and what you learned. Links to code, papers or a short write-up are welcome; published research is not required.
Email jobs@10xops.com with “Lead LLM Engineer” in the subject.