Feed

We’ve been prototyping typed primitives for retirements, vintages, and attestations. The goal is to make it trivial to build climate apps with strong provenance and composability.

We’re testing satellite-derived biomass + ground truth. Curious how teams handle drift, versioning, and audit trails—especially when models change month to month.

Looking for datasets that combine land surface temperature + impervious surface + socio-economic indicators. Ideally global or at least multi-city. What’s your go-to stack?

If you had to define a lightweight standard for reproducibility (data lineage, env pinning, artifacts, eval), what are the non-negotiables?

We’re debating between simple confidence intervals vs full distributions (and how to communicate it without overwhelming). What works best for reviewers and decision-makers?