Developer tooling at nCino for 30+ engineering teams

The summer

From May to September 2022 I was a DevOps Software Engineer Intern at nCino in Wilmington, NC. I sat on the platform side of the house, which meant my users were other engineers: more than 30 product teams who all depended on the same build and deploy path to ship.

That changed how I thought about the work. A product feature helps the people who use that feature. A pipeline change touches every team that builds on it, so a small improvement repeats across all of them, and so does a small mistake.

What I built

CI/CD pipelines and infrastructure-as-code. I built continuous integration and delivery pipelines and the infrastructure-as-code behind them, on AWS and GitHub Actions, for the 30+ product teams the platform supported. The goal was a shared path from commit to deploy that teams could use without each one maintaining its own.

Custom Datadog Lambda metrics. I built custom Datadog metrics for our AWS Lambda functions. They gave the company visibility into where Lambda spend was going, and that visibility cut cloud spend by about $100K a year.

A centralized developer hub. I launched a developer hub that lived in the build pipeline, so documentation was produced as part of the same process that built and shipped the code.

Bandit security scanning. I rolled out Bandit, a static analysis tool for Python security issues, across the organization through the shared pipeline.

Decisions

What I took away

The biggest lesson was that measurement comes before savings. Nobody could cut Lambda costs they couldn't see. The metrics themselves were not complicated; what mattered was putting the right numbers in front of the people who could act on them.

I also learned that platform work is mostly about adoption. Something only helps if teams use it, and they use it when it saves them effort on day one. That pushed me toward defaults that work out of the box and away from tools that need a long setup before they pay off.

Finally, working on shared infrastructure taught me to be careful. When a change runs in every team's pipeline, a mistake is everyone's problem, so I learned to make changes small, test them before they spread, and keep a way to roll back. I still use that habit in every project I work on.