Tag: machine-learning
2 entries matched this tag.
2026-03-2612 min readNote
⊘Sanitized
Supabase vs CloudKit for a local-first diagnostics app
An early working comparison of Supabase and CloudKit for a private, local-first iPhone diagnostics app. These are provisional findings and product-shape predictions that I want to revisit once the real build, sync path, and browser workflow are exercised.
2026-03-1813 min readNote
⊘Sanitized
Setgraph, Apple Health, TabPFN, and the day my lifting log got serious
I pulled my Setgraph history and Apple Health context into one table, rebuilt 1,548 exercise-sessions, benchmarked six tabular models, and learned three things fast: completed session quality was more explainable than I expected, cardio 24-48 hours before lifting looked better than expected, and personal-record windows were narrower than normal training defaults.