I’m refining a series of logistic-regression models in Python and would like hands-on guidance from someone who knows SciPy’s stats and optimize sub-modules inside out. My current notebook reaches acceptable accuracy, but I’m running into questions about link-function choices, solver stability, convergence diagnostics, and the best way to structure reusable helper functions. What I’d like from you • Review a short code sample (≈200 lines) that calls scipy.optimize.minimize for a custom log-likelihood. • Explain, with examples, how to improve numerical stability and interpret the Hessian output for standard-error estimation. • Suggest parameter-regularisation strategies available directly in SciPy or via light dependencies I can bolt on. • Join one or two live screen-share sessions (30-45 min each) so we can step through residual plots, goodness-of-fit tests, and any edge-case handling. All work will happen in a clean Python 3.11 environment with NumPy, SciPy 1.11, Pandas, and Matplotlib already installed, so no need for heavy-weight ML frameworks. Deliverables are the commented revisions to my script plus a concise summary of the changes and reasoning. If you’ve previously tuned logistic models in SciPy (not just scikit-learn) and enjoy explaining the “why” as much as the “how,” let’s talk.