A cloud migration finished on schedule with no release freeze for a German fintech
Two DevOps engineers took over the migration mid-flight after the lead left.
- Product
- Lending platform
- Technologies
- Kubernetes, Terraform, Azure, GitLab CI
- Timeframe
- 2025 – 2026
The migration had a deadline, a regulator and no owner — we supplied the owner.
The challenge
Their infrastructure lead resigned four weeks into a twelve-week migration. Half the services were containerised, the pipelines were half-rewritten and the audit date did not move.
Freezing releases was not an option: two product commitments were already announced to customers.
The audit date was fixed by the regulator, two product commitments were already public, and half the services were containerised — so the work could neither pause nor restart from scratch.
In a nutshell, when the company came to us it was grappling with:
- Infrastructure lead gone four weeks in.
- Pipelines half rewritten.
- A fixed regulatory audit date.
- Two release commitments already announced.
Key solutions
Taking over a cloud migration mid-flight
Two of our DevOps engineers picked the work up where it stood, finished the containerisation, rebuilt the pipelines with environment parity, and documented every step for the audit.
Releases continued through the whole migration on a blue-green setup the client's team now runs themselves.
Solutions we provided:
Continuity first
We picked the work up where it stood instead of restarting, so no announced release date moved.
Environment parity
Pipelines rebuilt so staging and production behave the same — the source of most of the pre-migration surprises.
Audit-ready documentation
Every step recorded as the regulator expects, handed over with the runbooks the client's team now uses.
Team on the project
Two engineers who owned the migration
- DevOps
- Platform engineering
Two DevOps engineers took over a migration already in flight, finished the containerisation, rebuilt the pipelines and documented everything the audit required.
The client's head of technology set priorities; our engineers ran the migration board and the audit evidence pack.
Fully remote, with one on-site week in Frankfurt for the cut-over weekend.
The migration was run remotely, with on-site presence where it counted:
- Cut-over weekend in Frankfurt
- Audit preparation sessionswith the client's compliance officer
- Weekly migration board
- Handover of runbooksto the client's platform team
- GermanyDE
- Saudi ArabiaSA
- PolandPL
The pipelines and runbooks now belong to the client's own team, who have run two further cut-overs without us.
Within days we had someone who owned the migration again. Nothing was announced late.
Results
- 12 weeks
- migration completed on time
- 0
- release freezes
- -35%
- infrastructure cost
- 100%
- audit evidence delivered
Within days we had someone who owned the migration again. Nothing was announced late.
Project completed:
A migration that stayed on schedule
The client went from a migration with no owner and a fixed audit date to a completed cloud move, rebuilt pipelines and an audit evidence pack — without freezing a single release.
Infrastructure cost fell 35%, the audit passed on the original date, and the client's platform team has since run two further cut-overs on their own.
Two DevOps engineers took over mid-flight, finished the containerisation, rebuilt the pipelines with environment parity and documented every step for the regulator.
Taking over someone else's half-finished migration is the hardest thing we do. Finishing it on the original date is the part I am proud of.