A senior engineer who owns and develops the platform behind the ARC-AGI series of benchmarks. This person will act as the technical owner and architect of our benchmark infrastructure, from stabilizing the current system to laying the foundation for future versions. This is a remote, full-time role.
What would you do:
- Stabilize and expand V3 backend and infrastructure – own performance to keep current benchmark platform reliable
- Build validation and testing layers – systems for automated model runs, scoring, reproducible eval pipelines, and capturing and querying data exhaust so the team can perform in-depth model analysis
- Support initial ARC-AGI-4 implementation by building the necessary backend and platform pieces for new environments, human data collection, scoring, and deployment
- Set the initial technical basis for ARC-AGI-5
What we are looking for:
- Strong backend engineering with Python, plus distributed systems, SQL, cloud infrastructure and production reliability experience
- Experience building evaluation harnesses, testing pipelines, experiment/data logging and analysis workflows – ideally for AI/ML systems or other high-volume technology platforms
- Senior enough to act as the technical owner and architect of the benchmark platform (we have a high agency team)
<a href