ML team standups should be different from regular engineering standups. Here's why ours changed.

Traditional standup: "Yesterday I worked on the feature extraction module. Today I'll continue. No blockers."

That tells the team almost nothing useful for ML work.

Our ML standup format:

"Yesterday I ran experiment X [link to W&B]. Hypothesis: adding temporal features would improve F1 by 5%. Result: F1 improved 2.3% but latency increased 40%. Today I'm investigating whether caching the temporal features resolves the latency. Decision needed by EOD: is 2.3% F1 improvement worth the added complexity?"

The key additions: experiment results with links, hypothesis and whether it was confirmed, and explicit decision points.

This changed our team dynamics completely. Instead of working in silos and presenting results at the end, we're making collective decisions throughout the process. Someone might say "I tried temporal features on a different project and caching solved the latency issue — here's how."

ML is inherently experimental. Your standup format should reflect that. If your standups sound like software engineering standups, you're losing valuable collaboration opportunities.