Thinking Machines Lab · posted · checked
Software Engineer, Evaluation Platform / Infra
PythonReactRustTypeScript$300k – $350kOn-site · New York and San Francisco
Who can apply
On-site · New York and San Francisco
Location: New York / San Francisco
ABOUT THINKING MACHINES
The mission of Thinking Machines is to build AI that extends human will and judgment. We are training frontier models with Inkling, developing Tinker to let people make models their own, and crafting interfaces that broaden human-AI communication. We believe the future worth building is human, and we're hiring people who want to build it.
ABOUT THE ROLE
Evaluation is one of the most important pillars of building frontier AI systems. It guides research direction, powers experimentation, and helps us understand whether changes to data and training are improving the capabilities and behaviors we care about.
To support this work, researchers need a powerful, self-serve platform that makes it easy to author evaluations, run them or reproduce them reliably at scale, and extract insight from the results. The platform must support both standardized external benchmarks and fast-moving internal evaluations, many kinds of tasks and graders, and inspection from aggregate metrics down to individual model trajectories.
In this role, you will design and build this platform end to end. You will work across Python frameworks, data pipelines, APIs, and user-facing applications, and collaborate closely with pre-training, post-training, and applied teams to improve how we evaluate models and turn results into research decisions.
WHAT YOU'LL DO
Design, build, and maintain the platform for authoring, running, tracking, and analyzing model evaluations that are critical in day to day work and model releases.
Work across evaluation libraries, distributed backend systems, data pipelines, APIs, and user-facing applications to deliver capabilities end to end.
Build flexible abstractions for evaluation tasks, environments, graders, datasets, and model outputs without constraining fast-moving research.
Make evaluation results reproducible and trustworthy through versioning, provenance, observability, failure recovery, and robust quality controls.
Partner directly with researchers to identify bottlenecks and turn bespoke workflows into self-serve systems that work across teams.
Work with Research Tooling, the engineering team behind Thinking Machines’ internal research platform.