Onboarding engineers into regulated environments in under two weeks
Access approvals, compliance training and a codebase nobody has documented. How we get placed engineers productive without cutting a single control.
Delivery Practice
Tvisha Scale Division
In a bank or a hospital, a new engineer's first two weeks are usually spent waiting: for a laptop, for an access approval, for a training module to be scheduled. Nobody defends this — it is simply what happens when onboarding is treated as a sequence of tickets rather than a designed process. It is also entirely fixable without weakening a single control.
Parallelise the paperwork
Background checks, compliance training and access requests do not depend on each other, and none of them depend on a start date. We initiate all three the moment an offer is accepted, with the client's security team briefed that the requests are coming. On a recent banking engagement this alone moved first-commit day from day fourteen to day four.
Give them a real first task
The traditional first week — read the documentation, watch the recordings — produces the illusion of preparation. We assign a genuine, small, low-blast-radius change that touches the build pipeline and the deployment path end to end. The engineer learns the system by moving through it, and the team learns whether the onboarding path is actually intact.
- Day 1–2: environment, access, a running local build
- Day 3–5: one small production change, paired, all the way to release
- Day 6–10: a bounded feature with review from the receiving team
- Day 30: a written checkpoint with the client lead
Pair, do not document
Most regulated codebases are under-documented and the documentation that exists is wrong in ways nobody has noticed. Ten days of pairing transfers more accurate context than any wiki, and it produces a by-product: the pair notices the wrong documentation and fixes it while it still matters to someone.
Keep reading
The code review comment that changed how our teams work
Four words in a pull request template did more for our review culture than any process document: "What made this hard?"
5 min readRetrofitting AI into legacy enterprise systems without a rewrite
Most enterprises do not need a greenfield platform to benefit from machine learning. A pragmatic pattern for wrapping models around the systems you cannot replace.
8 min readStaff augmentation or a dedicated development centre? A decision framework
Duration, domain depth and IP sensitivity decide this — not headcount. The four questions we ask every client before recommending a model.
6 min read