Start with the scientific problem.
We can discuss the question, available data, required deliverables, and what a defensible validation strategy should look like.
Useful context for an initial discussion
- Scientific or computational objective
- Available data or model outputs
- Current pipeline or evaluation method
- Expected deliverable or decision