Your DevOps engineer gave notice. Here's what to do this week.
When the only person who understands production leaves, their last two weeks matter more than the job posting. Digital Buttes runs knowledge-transfer sprints for exactly this situation: we sit with the departing engineer, capture what lives in their head, take over on-call, and run the platform until your team, or a new hire, can.
The knowledge-transfer checklist
Spend their remaining time on this list, in this order. Access first, because everything else can be recovered as long as you can still log in.
Access and accounts
- Root and owner logins for AWS or Google Cloud, and who else has them
- Billing contacts and payment methods on every cloud and SaaS account
- Domain registrar, DNS, and TLS certificates, including how they renew
- GitHub or GitLab organization ownership and CI/CD admin
- Where secrets actually live, and which ones are only on their laptop
- Admin on SSO, device management, monitoring, and paging tools
- SSH keys, VPN, bastions, and the break-glass procedure
How things run
- How a deploy happens end to end, and how to roll one back
- Where Terraform state lives and how a change gets applied
- How the cluster gets upgraded and when it was last done
- Backups: where they are and whether a restore has ever been tested
- What is monitored, what pages, and who gets paged
- The manual steps and workarounds nobody wrote down
- Recurring chores: renewals, rotations, invoices, cleanups
- The things they are worried about
How we step in
We do the writing. The departing engineer talks and demonstrates; we record, document, and verify by doing each thing ourselves before they leave.
Week one: capture
Structured sessions with the departing engineer, screen recordings of every procedure, an inventory of accounts and access, and rotation of every credential they held.
Weeks two to four: stabilize
We take on-call, turn the recordings into runbooks, fix the scariest thing on their worry list, and put the platform under version control if it is not already.
After: hand off or keep us
Train your developers on the day-to-day, help you hire and onboard a replacement if you want one, or keep us on as your fractional DevOps and IT team.
If they have already left
Harder, but not unusual. We start from the cloud console and the repositories, map what is actually running, and document as we go. The first three things are always the same.
- Confirm you hold the root and owner accounts, then rotate every credential the former engineer had
- Find where deploys and Terraform state live, and prove a deploy still works
- Point monitoring and paging at someone who will answer
Common questions
How fast can you start?
Tell us their last day and we work backwards from it. The capture sessions are the priority; everything else can follow.
Do we have to hire a replacement?
Not necessarily. Many teams keep us on part-time instead, and hire when there is a full-time job to fill.
What if there is no documentation at all?
That is the usual starting point. The checklist above is designed for it.
What if the departing engineer is not thrilled about this?
Short, structured sessions with a clear agenda, and we do all the writing. Most people are glad someone is taking it seriously.
Tell us their last day
Book a call this week. We'll work back from their last day and tell you what to capture first.