Stanimir Stoyanov, black and white portrait
R&D leadership, 11+ years

Stanimir Stoyanov

I lead engineering teams. Building has always been the part I enjoy most.

The short version

I started as a developer, took a lead role in 2015, and have been running engineering teams since: project management first, then engineering management, now R&D direction. The editor stayed open the whole time, so the decisions get made by someone who will live with the code afterwards.

What the work looks like

Warehouse hardware that has to be right every time: pick-and-put-to-light systems, connected weight scales, the firmware and the cloud behind both. ESP32 sensor fleets where the deliverable is a gate, not a demo. A self-hosted Postgres that answers full-text and vector questions with no vendor in the path. Small read-only APIs so a language model can cite a filing instead of inventing one.

More than once I have been brought in after a breach: work out what the exploit actually reached, close the door it came through, and write the runbook so the next response is shorter. That work is never solo — the network and cloud teams are in it with you, or it does not hold.

How I work

  • 01 Reproduce before you fix. A bug I cannot make happen on demand is a rumour. First a failing check, then the smallest change that turns it green.
  • 02 Design the seams, then the parts. Interfaces, data shapes and module boundaries are the decisions that outlive every implementation behind them. Settled in writing before the code, so five people can work in parallel without stepping on each other.
  • 03 Reproducible, and secure by default. Builds, migrations and tests run identically on a laptop and in CI. Pinned dependencies, least privilege, secrets out of the repository, nothing deployed by hand. Which is also what makes a change cheap to reverse.
  • 04 Prefer the simple tool. A database, a shell script and one binary beat a diagram of nine services. Complexity has to earn its place, and most of the time it does not.

Working together

Contract work, an integration nobody wants to own, a device that needs a better interface than the one it shipped with, or a team that wants someone senior in the room for a quarter. AI is part of how I get there: it drafts, then tests decide what ships.