We’ve been in the data infrastructure argument for twenty years.
The firm is what happens when that much memory shows up on an engagement with your name on it.
Patrick McFadin, the firm’s founder, was in the room when Spark was a research project at UC Berkeley and when Kafka was an internal LinkedIn tool. He knew the people building them, and was part of a lot of the decisions that became the systems your team is running now. He’s been a committer and PMC member on Apache Cassandra for more than a decade — long enough that those titles meant sitting in a chat room at midnight, not a line on a résumé.
The point of that history isn’t nostalgia. The point is memory. When a team asks us whether a particular architectural bet is safe, we usually know who made that bet originally, what they knew at the time, and what they hoped would happen. We know which benchmarks were honest and which ones were gamed. We know which vendors quietly walked back their claims two years later, because we were in the conversations when the claims were made.
McFadin Data and AI Advisory is what happens when twenty years of those conversations — the ones you can’t Google and the ones that never made it into a whitepaper — show up on an engagement with your name on it. We tell you what we think. We credit the people who taught us to think it. And we stay long enough for your team to know it too.
The founder
Patrick McFadin. Founder.
Patrick started his career as an Oracle DBA in the late 1990s, back when “production database” meant a room full of blinking lights and a pager that went off at 3 a.m. He spent fifteen years running databases for real companies with real money on the line, eventually climbing into chief architect and CTO seats at a string of internet-era firms.
In 2012 he joined DataStax, where he spent thirteen years working on Apache Cassandra as a committer, then a PMC member, then an Apache Software Foundation member. He served as VP and Principal at DataStax, helping hundreds of companies run Cassandra at scale: migrations off Oracle and DB2, moves onto Kubernetes, production incidents, cost fires. Along the way he was around for the formative moments of the distributed data generation — Spark when it was still a research project at UC Berkeley, Kafka when it was an internal LinkedIn tool — and knew the people building them. He co-wrote Managing Cloud Native Data on Kubernetes (O’Reilly, 2022) with Jeff Carpenter. He gave the talks at KubeCon, ApacheCon, QCon, and Strata. He reviewed more than a hundred production data architectures across finance, retail, SaaS, and AI platforms.
In 2026 he founded McFadin Data and AI Advisory to do the work directly.
Apache Cassandra PMC O’Reilly Author 20+ years in Silicon Valley data technology 50+ conference talks Making teams successful from pre-seed to Fortune 500
How we work
- 01
We’ve been in the room.
Twenty years of architecture decisions, benchmark debates, and vendor claims — we were in the conversations when most of them got made. That’s the memory we bring to your problem.
- 02
We tell you what we actually think.
No hedging, no vendor politics, no deck-safe consensus. If an architecture is wrong, we say so. If the honest answer is “you don’t have this problem,” we say that too.
- 03
We credit the people who taught us.
The knowledge we bring is built on a lot of other people’s work. We name them, we point at their writing, and we send you to them when they’re the right answer.
- 04
We stay long enough to be useful.
A report that lands in a drawer is not the point. We stay on the engagement long enough for your team to understand what we changed and why, and to do it themselves next time.
- 05
We don’t hand off to juniors.
Every engagement is led by someone who has shipped production data systems for decades. If that person can’t take the work, we say no.
Let’s look at it together.
Bring us whatever you’re wrestling with. Thirty minutes, no pitch. We’ll tell you what we see.