Role DescriptionYou will work across backend systems, product surfaces, data pipelines, internal tools, evaluation workflows, and customer-facing prototypes.
You might build Forge, the platform that imports traces, logs, recordings, and schemas; the interfaces that let humans tune scenarios, perturbations, and rewards; the services that turn raw streams into executable environments; or the tools researchers use to inspect rollouts and model failures.
This is not conventional product engineering. You will be building software in the middle of a fast-moving research loop, where the product surface, backend systems, data model, and customer workflow often evolve together.
You Will Work On- Develop product surfaces for researchers, domain experts, and customer teams to inspect, tune, replay, and validate generated environments.
- Build backend systems for trace ingestion, schema handling, environment generation, task generation, scoring, and telemetry.
- Create internal tools that help researchers move faster across evals, rollouts, verifiers, rewards, and failure analysis.
- Turn customer workflows into robust software systems that can be reused across frontier labs and enterprise deployments.
- Ship pragmatic, high-quality software in a fast-moving, deeply technical team.
What We're Looking ForWe're looking for someone who is excited to build software at the boundary of product, infrastructure, and AI research.
You may be a strong fit if you:
- Have strong product engineering taste and can build clear interfaces for complex technical workflows.
- Are comfortable working across backend systems, data pipelines, product surfaces, and internal tools.
- Can turn ambiguous customer or research workflows into robust, reusable software.
- Care about correctness, usability, observability, and iteration speed.
- Want to build systems that are part of the core training loop, not a wrapper around it.
