Personal blog on early retirement, selling tradelines, and analytics projects. I write about what I learn as I am learning it.
analytics
The app was working. On my laptop. Which is the same as not working, for most purposes.
If the goal was to keep it to myself — test it occasionally, tinker with it, update the model when I felt like it — a working local copy would be enough. But that’s not what I built it for. The point of converting these models to web apps is that they can run anywhere, for anyone, without requiring someone to have Python installed and know how to use a terminal.
So: deployment.
At some point during this project I stepped back and looked at what we’d built.
Three files. A working web application. Deployable for $7 a month. Zero web development experience going in.
That felt worth writing down.
When you convert an optimization model from Excel to standalone Python, you expect to do some work. You expect to rewrite the data loading, restructure the variable definitions, test the output. What you don’t expect is for the model to fail in three distinct ways, each one caused by something the Excel version was handling silently without you knowing it.
That’s what happened here. Three bugs. All real. All the kind that would have stayed invisible forever if the model had stayed in the spreadsheet.
Continue reading “Three Bugs That Would Have Stayed Hidden in Excel”PuLP is a Python library for writing optimization models. MILP stands for Mixed Integer Linear Program. CBC is an open-source solver. Together, they’re what makes the Staff Scheduler work — and together, they represent something I find genuinely interesting: the fact that problems that used to require expensive commercial software and specialized hardware can now be solved on a laptop, for free, in a few seconds.
Let me explain what’s actually happening under the hood.
Continue reading “PuLP, MILP, and CBC: The Alphabet Soup Behind Your Schedule”Years ago, I was doing gigs on Fiverr. Mostly Excel work — macros, pivot tables, solver models. I had a profile and a handful of listings, and I’d written about that whole love/hate experience back in 2017 (short version: the clients who treat you like a vending machine are the price you pay for the ones who bring you genuinely interesting problems). One day, a genuinely interesting problem showed up.
A training company needed help scheduling their staff. Multiple clients, multiple courses, multiple instructors, limited hours, shared equipment, weird simultaneous constraints. Classic Operations Research territory. I said yes.
Continue reading “The Client Who Changed His Mind (And Why I Finished the Job Anyway)”