Research · Case Studies · STC (SCIENCE & TECHNOLOGY CORP.)

CASE STUDY · FEDERAL · CI/CD & DEVOPS

STC's satellite scientists went from manual deploys to one-click production.

Aeroxis built and ran the GitLab CI/CD pipeline for STC (Science & Technology Corp.), the federal contractor behind NOAA and NASA satellite programs. We took a Python pipeline that processes GOES ABI satellite imagery from manual, error-prone deploys to automated build, test, and continuous delivery across dev, staging, and production on AWS EKS — so research scientists could ship their own work, safely, without a DevOps engineer in the loop.

CLIENT

STC (Science & Technology Corp.)

CATEGORY

Federal · CI/CD & DevOps

PRODUCTION DEPLOYS

Manual → 1-click

LIVE ENVIRONMENTS ON EKS (DEV · STAGING · PROD)

3

Aeroxis built and ran the GitLab CI/CD pipeline for STC (Science & Technology Corp.), the federal contractor behind NOAA and NASA satellite programs. We took a Python pipeline that processes GOES ABI satellite imagery from manual, error-prone deploys to automated build, test, and continuous delivery across dev, staging, and production on AWS EKS — so research scientists could ship their own work, safely, without a DevOps engineer in the loop.

01What changed for STC

By the end, STC's research scientists were shipping their own satellite-data software to production with a single click — no DevOps engineer standing between their work and the mission. That was the goal; that's the outcome. Here's what it replaced, and how.

02The starting point

STC (Science & Technology Corp.) — the federal contractor behind engineering and science support for NOAA and NASA satellite programs — had scientists running a Python ETL application that turned raw GOES ABI satellite imagery into analysis-ready data products. The science was solid; delivery was fragile. Deploys were manual — SSH in, copy files, restart, hope. No CI meant nothing built or tested the code automatically, so regressions reached production. No repeatable environments meant “works on my laptop” was the plan. And these were research scientists — delivery plumbing was never their job.

03The pipeline: build, test, ship

We built the whole delivery spine on GitLab — created the organization and repositories, then a CI pipeline that ran on every push: unit and integration tests, linting and code-quality gates, dependency and container security scanning, and a Docker image build pushed to AWS ECR. Nothing reached an environment that hadn't passed the gates.

04Kubernetes on EKS, and a deliberate call

Deployments targeted Kubernetes on AWS EKS, pulling tagged images from ECR. We defined workloads as explicit kubectl manifests rather than Helm charts — on purpose. Helm 3 was under a year old when the project began, and you don't bet a production science pipeline on year-old tooling. Explicit manifests also let a scientist read a git diff and see exactly what would change in the cluster: auditability over abstraction.

05Automated where it's safe, human where it counts

Merges deployed automatically to dev and staging. Production required a manual approval — a deliberate gate, not an oversight. It's how we still think: automation is fast and tireless; a human is the trustworthy checkpoint. The pipeline did the fast, repeatable work and stopped at the door of production for a person to say “go.”

06Mission context

STC (Science & Technology Corp.) is the federal contractor behind engineering and science support for NOAA and NASA satellite programs. The pipeline processed GOES ABI weather-satellite imagery into analysis-ready data products. Delivered by Aeroxis, 2020–2022.

MEASURED OUTCOMES

Manual → 1-click

PRODUCTION DEPLOYS

3

LIVE ENVIRONMENTS ON EKS (DEV · STAGING · PROD)

2020–2022

IN PRODUCTION, ~2 YEARS

WORK LIKE THIS

Think we could help you
ship the thing
that's been stuck?

Engagements typically start with a two-week fixed-fee diagnostic — fast, narrow, ends with a plan you can execute with or without us.