02 — Story
Four chapters.
How a curious kid with a slow laptop ended up shipping AI for enterprise floors.
Chapter 01 · 2019
It started with a slow laptop.
I learned to code because the machine I had couldn't run anything heavy — so I had to understand what every line cost. That constraint became the habit: build small, measure, then grow.
First build2019
Chapter 02 · 2023 — 2025
Enterprise taught me the unglamorous half.
At AVASOFT and Zeb I shipped into environments where a failed retry is a phone call at 2am. Logging, idempotency, handoffs, rollback plans — the parts no demo shows.
On the floorAVASOFT
Chapter 03 · Now
Now I make AI behave itself.
Agents with guardrails. Retrieval that cites its source. Pipelines that page me before a client notices. Generative AI is only useful when it is boring in production.
Orchestration2026
Chapter 04 · Next
Looking for the next hard problem.
Fixed scope, honest estimates, documentation you can actually read. If your AI feature is stuck between a demo and production, that's the gap I work in.
Next buildOpen
