Atlas Digital Engineering is the contracting practice of James Griffis, a senior DevSecOps and platform engineer. It exists to put that experience to work for teams who need their platforms to be fast, secure and understood by the people running them.
Background#
James has worked in the field since 2011, starting in backend development and datacentre hardware and moving through systems engineering into what is now called DevSecOps. That path matters: it means speaking developer, operator and security in the same conversation, which is usually where platform work succeeds or fails.
Over the last decade that has meant senior roles across banking, fintech, payments, professional services, health and large enterprise, often as the engineer brought in to make Kubernetes, CI/CD and cloud security actually work for the teams using them. Along the way: an Alibaba Cloud MVP, an Advanced Diploma in Computer Systems Engineering, and hands-on datacentre build work.
How we work#
- Evidence over assumption. When something is broken, gather the data point before the invasive fix. Byte-for-byte verification against a running system beats “looks plausible”.
- GitOps by default. Issue, branch, pull request, review. The issue thread is the record of decisions, and documentation is fixed in the same change that made it wrong.
- Signed everything. Every commit is GPG-signed from a hardware key. Supply-chain integrity starts at home.
- Straight talk. Failures, skipped steps and open gaps are reported plainly. Overselling a fix costs more than admitting the caveat.
- Leave the team stronger. Simple solutions to complicated problems, less red tape, and a team that is more self-sufficient at the end than at the start.
Lab and open source#
Atlas runs its own infrastructure: a blade chassis, TrueNAS storage, and a private cloud using OpenStack. It is the testbed for driving creative solutions and reinforcing our solutions testing, which means new approaches have already been run against real hardware and real failure modes before they reach expensive cloud infrastructure.
Tooling that comes out of that work is published where it can be. TrueNasCinder, a Python client and OpenStack Cinder driver for TrueNAS Scale, is on the Atlas GitHub.