Field NotesLab notes from the network.
Honest essays from researchers about what's working, what isn't, and what we're learning. We publish negative results, retrospectives, and field reports — not just the wins.
What we got wrong about Kinyarwanda tone in v0.3
A retrospective on tone modeling, why our F1 collapsed in Eastern Province, and the dataset patch that fixed it.
Read the full post →Notes from a week with cassava farmers in the Ashanti region
Field deployments humble models fast. Fourteen farmers, two cooperatives, one growing season.
Read the full post →Why our antenatal model intentionally returns 'unsure'
Calibration matters more than accuracy in clinical decision support.
Read the full post →Sahel drought forecasting at 10 days: what's actually possible
What's signal, what's noise, and where the next gains will come from.
Read the full post →Open-call: 2026 fellowship applications now live
What we look for, how we score, and how to make a strong application.
Read the full post →Edge-first ML: shipping a 14MB model that runs everywhere
Quantization, pruning, and the unsexy systems work behind a model that runs on a $40 phone.
Read the full post →Reproducibility as a deliverable, not a footnote
If we can't re-run your training in a notebook six months later, it didn't happen.
Read the full post →Community-owned data governance, in practice
Notes from the FAccT paper, but for practitioners.
Read the full post →Subscribe to the monthly digest
One email per month. New papers, deployment notes, and open calls. No tracking.
Subscribe