AI Agent Evaluation Architect

Mindrift in Illapel, Chile, offers an opportunity to build a dataset for evaluating AI coding agents. You will develop complex tasks and evaluation criteria within simulated environments that mimic real-world development scenarios.

Your contributions will include creating tasks for an AI’s coding capabilities, writing tests to ensure correctness, and analyzing agent performance. This position focuses on project-based work and requires collaboration with AI tools to challenge AI models effectively.