Handover

We leave.
Your team keeps
shipping safely.

Workshops and runbooks for the people who will publish corpora, tune agents, and interpret eval failures, free of slide decks for executives who will never SSH.

The worst outcome of a successful pilot: Corvux builds something brilliant and your team is scared to touch it. Ops training is the deliberate handover, publish a corpus, run an eval, roll back a bad agent, with us in the room until muscle memory sticks.

Workshop modules

Hands on keyboard.

Day 1
Platform tour
Models, routing, logs, backup, where Kenya residency lives in your stack.
Day 1
Corpus lifecycle
Ingest, chunk, re-index, rollback, on a sandbox doc set from your sector.
Day 2
Agent publish
Build in Studio, pass eval, get approver sign-off, end to end.
Day 2
Incident drill
Simulated jailbreak + PII leak, execute runbook, notify DPO, patch prompt.
$ corvux train verify --cohort ops-ke-2026-q2

PUBLISH 3/3 agents passed supervised exercise

EVAL red-team drill completed · 42 min MTTR

RUNBOOK signed off by platform lead

✓ handover gate cleared
2+2
Days + clinics
Sandbox
Safe mistakes
Runbooks
Yours to keep
Nairobi
On-site option

Schedule training the week before go-live expansion, ahead of a six-month delay when the one person who understood the system left for a fintech.

FAQ

HR and platform leads.

Who should attend?

Platform engineers, ops owners, and at least one risk/compliance observer, a small group, short of the whole company.

On-prem or managed?

Curriculum adapts to your deployment: appliance racks, managed API, or hybrid.

How long?

Typical: 2-day intensive + 2 follow-up clinics. Custom tracks for contact-center ops vs ML platform.

Certification?

Corvux completion record for your HR file, an HR-file record, short of an industry licence. Competence is demonstrated in supervised publish exercises.

Book a handover cohort

Tie training to a real go-live date. Sandbox exercises use your architecture, your own architecture, off a demo tenant.

Plan ops training