Platform engineering sounds like a big-company thing. But I think every startup past 20 engineers needs a small platform function. Here's why.

The problem platform engineering solves

At 5 engineers, everybody knows how to deploy. At 20, they don't. People start copy-pasting deployment configs, breaking things, and asking the same questions in Slack every day.

You need one person who owns 'how we ship software,' even part-time. That person is a platform engineer whether you call them that or not.

What a startup platform function does

1. Owns the deployment path. One golden path from git push to production. Documented, maintained, and defended from one-off exceptions.

2. Owns the dev environment. Laptop setup, local testing, shared services. New hire productive in days, not weeks.

3. Owns the shared services. Auth, logging, tracing, secrets management. Used by everyone, owned by no one until you assign it.

4. Owns the developer experience. CI speed. Local/prod parity. Error messages. The stuff that's no one's job but costs everyone.

What it doesn't do

It doesn't build a K8s abstraction layer that rivals AWS. Startups can't afford that. Use off-the-shelf. Customize lightly.

When to start

At 15-20 engineers, if you're still asking 'why is my build failing' in Slack 3 times a week, you're ready.

Before 15, just have engineers fix each other's stuff and take turns being the 'ops person.' It's not efficient, but it's cheaper than a full function.

The hire

Hire someone who loves developer experience. Not the best infrastructure engineer on the team the one who genuinely cares about making everyone else faster. That's a different skill set.

Written by Dr. Samson Tanimawo
BSc · MSc · MBA · PhD
Founder & CEO, Nova AI Ops. https://novaaiops.com