Lead Engineer, AI Platform

About the role

ABOUT US

Circle is building the world’s leading AI-powered, all-in-one platform for digital businesses. We make it possible for creators, coaches, educators, and businesses to bring together their audience with engaging discussions, live streams, events, chat, courses, and payments — all in one place, all under their own brand.

We’re proud to be a fully remote company of around 270 (and growing!) team members from 30+ countries around the world. We seek exceptional individuals around the world, set them up to do the best work of their lives, and in turn, create a meaningful impact in their own lives. We don't track hours, but we do manage for high expectations very closely. We collaborate across time zones, are highly async, and like to document a lot.

Every year, we bring the whole company together in beautiful places around the world for our company offsites. So far, we’ve hosted offsites in Turkey, Portugal, Mexico, Thailand, Colombia, Italy, Ireland, and more, with still more to come!

Check out our Careers https://careers.circle.so/ page for more about working at Circle.

ABOUT THE ROLE

The AI Quality engineering team at Circle owns the foundation for measuring, diagnosing, and improving the quality of Circle's AI-powered features. This team focuses on building the infrastructure to measure, diagnose, and improve production AI systems, rather than ML research or model training.

We're looking for a Lead Engineer to help us build out the evaluation frameworks, observability tooling, and diagnostic infrastructure that tell us whether our AI Agents are working well, where to improve them, and how to make them faster and more cost-efficient.

This is a hands-on player-coach role. You’ll also lead and manage the AI Quality engineering team, with an ambitious and growing roadmap.

If you're excited about making AI systems work reliably, efficiently, and at scale, this is for you.

WHAT YOU'LL BE DOING

- Build and own our evaluation infrastructure. Design the CI/CD pipelines, scorers, and datasets that tell us whether Circle's AI agents (planners, tool-callers, and sub-agents) are actually working, from a single tool call to a full multi-turn conversation.

- Diagnose exactly where quality breaks down. Trace failures across the agent pipeline including plan creation vs. execution, tool selection, tool trajectory in complex areas like workflows, site builder, and analytics. Turn what you find into prototypes that solve the issue or identify priorities for AI core engineering can help.

- Grow the datasets that make evaluation possible. Stand up annotation workflows and build out AI-generated and simulated conversations so we can cover more of the product faster than manual labeling alone.

- Run structured experiments across prompts, models, and the agent harness. Evaluate new and open-source models against our production baseline, build the framework we use to decide when to shift models, and chase cost and latency wins through model swaps, caching, and routing by plan complexity.