Best Practices Catalog
Best practices for scientific workflows, emphasizing how AI-enabled and agentic systems should be designed, evaluated, and documented in research settings. A starting point for workflow design guidance instead of building conventions from scratch — for research software teams, AI-for-science teams, and workflow designers.
AI Powered Data Pipelines
Incorporating AI methods into scientific data processing workflows; where automation, learned components, or agentic assistance can improve pipeline design. For data engineering teams, workflow developers, and AI-for-science practitioners.
AI Evaluation Tutorial
Evaluation methods for AI systems: quality, reliability, and scientific relevance when assessing AI-enabled workflows. For evaluation teams and developers integrating AI into scientific processes.
Best Practices for Designing Evaluation of Agentic AI for Science
Evaluation design for agentic AI: framing tasks, defining metrics, and reasoning about success in realistic scientific settings. For agent developers, evaluation leads, and science-facing AI teams.