Analogy AI is building the data infrastructure behind self-evolving AI systems, focusing on automating the full lifecycle of data. The role involves leading AI research and engineering efforts to create scalable systems that enhance data loops and agentic systems through collaboration and prototyping.
Responsibilities:
- Lead novel research on agentic systems and self-evolving data loops (offline evals + production signals → next data batch)
- Work with leading customers to design and curate high-quality data and standardize the data curation pipelines
- Design and scale environments for agents (synthetic + real-world, tool-use, long-horizon tasks, multi-step reasoning, multimodal where useful)
- Build training + evaluation pipelines: metrics, rubrics, automated judging, failure clustering, regression tests, and dashboards
- Develop data generation and curation systems: schema design, quality filters, safety checks, dedup, provenance, and preference signals
- Explore environment scaling methods (curricula, task generation, distribution shaping, difficulty calibration, reward design)
- Prototype quickly (tight iteration loops), then harden into production-grade components
- Collaborate across product/engineering to ship features that customers actually use