Research Engineer

present of AI; San Francisco (CA); On-site; USD 150,000 - 190,000; Full time

Job summary

TrainLoop is a post-training research and product lab that trains specialized models for long-horizon tasks, per the company.

A research engineer in this setting turns research ideas into working training and evaluation pipelines, balancing research and product goals in an environment that ships both.

Responsibilities

• Turn research ideas into working training and evaluation pipelines for long horizon tasks.
• Code, train, evaluate models, and run careful comparisons.

Job description

A research engineer in this setting typically turns research ideas into working training and evaluation pipelines for long horizon tasks. A strong candidate brings solid coding, comfort with model training and experimentation, and the patience to run careful comparisons. Since the company builds specialized models and developer tools, the work likely sits close to both the research and the product.

How it fits the AI era

AI has made model training and evaluation faster to iterate on, but choosing what to measure and why still rests with people. This kind of role shifts toward designing experiments, reading results critically, and turning findings into reliable systems. The human part is judgment about what matters and what a result actually shows.

How it builds AI experience

Someone in this role would gain hands on experience with post training methods, long horizon task design, and evaluation of specialized models. That experience comes from running real training runs, debugging them, and building tools that others can reuse. Working at a lab that ships both research and product means the AI experience is practical rather than only theoretical.

Worth showing

• Evaluation methods
• Research to product

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About TrainLoop

TrainLoop is a post-training research and product lab that trains specialized models for long-horizon tasks, per the company.