McGinnis, Will

The first time I tried adding Bandit to our CI pipeline, I broke every deployment for a week.

Last fall I used AI agents, a mix of claude code and codex, to write a custom personal CRM in a language I don’t know: Swift.

I keep a list of restaurants I think people should try.

In our previous posts, we explored the Kelly Criterion and Fractional Kelly strategies.

Six months after I adopted Bandit, I realized we had a false positive problem.

Atlanta AI Dinner has been going for a while now, and we’ve had some really great events over the past year.

In our previous post on the Kelly Criterion, we explored the mathematically optimal strategy for maximizing long-term bankroll growth.

I’ve written about frameworks that help you respond when a situation unfolds - PDCA for systematic improvement, ASPIRE for real-time clinical judgment, RPD for split-second pattern matching.

We’re currently doing some construction on our house, which means we’ve been living in an Airbnb for a few months with our three-year-old.

If you’ve spent any time around systems engineers, you’ve probably heard the terms “verification” and “validation” thrown around with an almost religious reverence.

Over Christmas break, I also built an iOS app called Bet A Day.

I ride bikes a lot.

Every summer my family goes to North West Michigan to visit family and escape the summer heat in Atlanta.

Coldwater Mountain has earned its reputation as one of the Southeast’s premier mountain biking destinations.

Every year I try to find a theme.

I find it genuinely interesting to see where things I write end up.

2025 was a year of transitions, but not always change.

I’ve been using ContextSwitch to manage my todos for a while now, but the new diary feature has made it much more valuable.

I build software that incorporates AI, which means I spend a lot of time thinking about what these tools can and can’t do.

I’ve been building out a collection of guides on this site that are a bit different from regular blog posts.

The new century brought new beginnings.

Charles McGinnis-“Old Charles,” as we’ll call him to distinguish him from the many Charleses who would follow-was probably born about 1752, possibly in Pennsylvania.* 1752 - Benjamin Franklin’s kite experiment with electricity He is the patriarch of our l…

Imagine standing on the docks at Belfast or Newry, looking out at a wooden ship that would be your home for the next two months.~ We don’t know exactly which port our McGinnis ancestor sailed from, but Belfast and Newry were the busiest for Ulster emigran…

The journey of the McGinnis family to America did not begin in Ireland.

🦄 Izzy Mode 🦄 🌈 ⭐ 🎀 🌸 ✨ 🎵🦄🎶 Welcome to Izzy's Music!

As 2025 winds down, I wanted to take a moment to reflect on what’s been a transformative year for PyGeoHash.

🦄 Izzy Mode 🦄 🌈 ⭐ 🎀 🌸 ✨ 🦄🌈✨ Welcome to Izzy's Bookshelf!

A collection of blogs and personal websites I enjoy reading.

A collection of aphorisms, quotes, creeds, or pithy phrases I’ve come across or heard and find worth remembering for one reason or another.

I’ve been building a lot of native macOS apps lately using AI pair programmers.

I got tired of manually creating header images for blog posts.

When OpenAI announced “deep research” in early 2025, it sounded like marketing speak.

I’ve been thinking about AI agents as tools that help humans work faster.

In my previous post I covered three Python frameworks for getting structured data from LLMs: instructor, outlines, and pydantic-ai.

Most enterprise uses of LLMs require structured output.

The beating heart of the internet has also been people just sorta putting things out there.

Last year I wrote about decision making under uncertainty, exploring how expected value calculations break down when facing real-world complexity.

I spent $1.00 and a Saturday morning testing prompt caching with Anthropic and OpenAI.

Anthropic released MCPB (MCP Bundles), a packaging format that lets you install MCP servers with a single click, similar to how Chrome extensions or VS Code plugins work.

Two papers dropped in the last month that should make anyone deploying AI agents nervous.

Annie Duke talks about a concept called “resulting” in her book Thinking in Bets: the tendency to judge decisions by their outcomes rather than their quality at the time they were made.

We’ve explored how systematic frameworks like PDCA and the nursing decision cycle enable structured decision-making even under pressure.

Making Better Decisions, Systematically I’ve been collecting decision-making frameworks for years, partly out of professional necessity and partly because I find it genuinely interesting how different fields have solved the same fundamental problem: how do

Fall has been a season of decision-making and risk for me, both in what I’ve been reading and thinking about.

Throughout this series on decision making structures, we’ve explored how different industries approach high-stakes decision making under uncertainty.

I’m excited to share that I’ve joined Scale Venture Partners as Chief Data Scientist.

I’m excited to announce Keeks 0.3.0, which introduces a powerful new betting strategy based on Robert Merton’s portfolio optimization work.

I built two projects over the past year or so that will probably never see production.

I keep finding myself typing the same prompts over and over.

A month ago I built Evergreen, a personal CRM in Swift despite never having written Swift before.