← All work
Accenture / Workflow automation
Approval workflow
One workflow. A human yes at every step.
End-to-end Linux patching: repo sync, prechecks, unsnooze, patch and reboot, tied to the change schedule, with Teams approval at each step.
3–4→1engineers on a window
5–6h→<1hhands-on time
Role
I built the end-to-end workflow on Accenture's AIOps platform.
Stack
Python, AIOps workflow platform, SUSE Manager, ServiceNow, Microsoft Teams
The problem
A change window took three or four engineers, each hands-on for five or six hours, across a window of six to eight hours.
What I built
- Five stages in order: repo sync, prechecks, unsnooze, apply patches, reboot, with a notice of what will be installed before it happens.
- Tied to the ServiceNow change schedule, not an ad hoc list.
- Every stage pauses for a human check through the Teams chatbot.
- An operator can hold the workflow so it does not advance past the current step.
Where it stands
Proven on non-production so far. Production and clusters are next. One person now monitors instead of three or four working by hand.
Accenture. Proven on non-production only so far. Production and clusters are next. Film is a reconstruction.