From spreadsheets to production AI.
Not a career pivot. A compounding of the same skill, applied to a harder problem.
I was an accountant who couldn't stop asking why the systems didn't talk to each other.
The ERP that didn't sync to the GL. The reconciliation that took 3 days because two teams used different keys for the same entity. The meeting where no one could answer a simple cash-flow question because the data lived in 4 places.
The insight wasn't technical. It was structural. The systems had no shared language. Requirements documents were about giving them one.
Requirements documents were my superpower. I just didn't know it yet.
A BA decomposes ambiguity into structure. Defines what "done" looks like before a line of code is written. Maps data flows, exception paths, edge cases. Translates between what a business needs and what a system can do.
When large language models arrived, the gap they needed filled was exactly this: clear input specification, defined output contracts, structured exception handling. The job title changed. The skill set didn't.
The first time an AI agent deployed code I wrote, I stayed up until 3am watching it.
Not watching it fail. Watching it work. Watching the deploy pipeline complete, the health checks pass, the Telegram notification arrive. A system that I had specified, the same way I would specify a business requirement, running in production without me touching it again.
That was the turn. Not a skill I had to acquire — a context where the skill finally fit.
The BA skillset IS the advantage.
They said you can't build production AI systems without a CS degree. This site is my response.
Not a defense. A demonstration. The ten pages you've just read aren't a portfolio in the traditional sense — they are a working argument. Each one is a system in production. Each one was specified before it was built.
The degree they meant was a proxy for structured thinking. I have that. I got it from five years of requirements documents, process maps, and stakeholder workshops.
Don't take my word for it.
The evidence —
- Homepage The index of everything that follows.
- Trading A live BTC dashboard built on CCXT and Supabase.
- Architecture The interactive system diagram that runs the swarm.
- Swarm The agent roster. Real services, live status.
- CX Swarm Enterprise customer experience, automated end-to-end.
- Lessons 41 field journal entries from building in production.
- Ideation A sunrise engine for daily structured thinking.
- Observability The live health dashboard for the swarm itself.