Writing from bizzib.ai: how things got built, what broke, and what the data said.https://chellecode.com/https://chellecode.com/writing/does-your-step-counter-undercount-women/https://chellecode.com/writing/does-your-step-counter-undercount-women/A Bayesian statistics audit of whether step counters miss more of women's steps. Short answer: if they do, it's small.Wed, 03 Jun 2026 00:00:00 GMTbayesianfairnesswearablesschoolhttps://chellecode.com/writing/teaching-a-random-forest-to-tell-walking-from-running/https://chellecode.com/writing/teaching-a-random-forest-to-tell-walking-from-running/It went from 56 hand-built features to 240 and got 86% right. The fun part is seeing which moves it still mixes up, and why. (Classical computer vision, random forest, SHAP.)Fri, 08 May 2026 00:00:00 GMTcomputer-visionexplainabilityschoolhttps://chellecode.com/writing/let-the-user-edit/https://chellecode.com/writing/let-the-user-edit/846 ghost edits, a plus-or-minus sign nobody warned me about, and one check counted twice. (How a read-only dashboard became a full-stack app.)Tue, 17 Mar 2026 00:00:00 GMTbizzibdashboardsserverlesshttps://chellecode.com/writing/automating-tax-prep-for-a-construction-business/https://chellecode.com/writing/automating-tax-prep-for-a-construction-business/Eleven accounts, 1,310 transactions, every Venmo payment counted twice, and an accountant waiting. (A three-stage Python pipeline and a review dashboard.)Wed, 04 Mar 2026 00:00:00 GMTbizzibdata-pipelinesdashboards