Research Engineer, Post-Training and RL
Train and evaluate the models behind our agents, from fine-tuning to reinforcement learning on loan outcomes.
About the role
We sit on something most AI teams do not have: expert corrections and real outcomes on every loan. Your job is to turn that into models that are measurably better at origination than anything off the shelf.
What you will do
- Fine-tune and post-train models on expert-reviewed origination data
- Design reward signals and RL setups grounded in real loan outcomes
- Build the benchmarks that tell us when our models beat the frontier on our tasks
- Make training and evaluation reproducible, cheap, and fast to repeat
- Publish or open-source what we can
What you bring
- Hands-on experience with fine-tuning, RLHF, RLAIF, DPO, or related methods
- Strong Python and experience with modern training stacks
- Rigor about evaluation: you distrust a number until you know how it was made
- Experience with document understanding or structured extraction is a plus
- A bias toward results that ship, not just results that publish
Working with the agents
You make the models underneath the agents better at this one domain than any general model can be.
Apply for this role
Three things to get started. A person reads every application, and we will reply either way.
Loan Labs is an equal opportunity employer. We consider every applicant without regard to race, color, religion, sex, sexual orientation, gender identity, national origin, age, disability, veteran status, or any other characteristic protected by law. If you need an accommodation to apply, write to careers@loanlabs.ai.
Other open roles on this team