Launching an AI data programme of more than 250 people at SLA from day one
A large engagement that had to perform to contract standards immediately, with no warm-up period. Every SLA commitment was met from the first day.
- My role
- Associate Project Manager
- Context
- Large-scale AI data engagement · one of three concurrent projects
- Scale
- More than 250 team members
- Day 1every SLA commitment met
- 250+team members at launch
- 100%SLA adherence across the portfolio
- 7business functions coordinated
Overview
A high-stakes engagement in AI data operations, staffed by more than 250 people. It sat inside a portfolio of three projects I manage, with 320+ team members in total.
The challenge
At this size, launch readiness has several moving parts that all have to land together: the workforce sized correctly, people trained and calibrated, access and tooling in place, team leads assigned, and cover for absences. Any gap appears in the first week’s numbers, and the client reads the first week as a signal of the whole engagement.
My role
As Associate Project Manager I owned end-to-end delivery and the coordination between Operations and the functions that support it.
Approach
- Capacity planning first. I sized the team against expected volume and quality targets before launch, so the plan didn’t depend on hoping the numbers would work out.
- Resource allocation. I structured the workforce under team leads so every part of the work had a clear owner and a clear target.
- Risk managed early. I identified launch risks in advance and agreed mitigations, including contingency staffing, before they could hit delivery.
- One plan, seven functions. I kept Finance, Workforce Management, Analytics, Quality, Training, IT and Operations working to the same plan and timeline.
Outcome
The engagement met all of its SLA commitments from Day 1. Across the three projects I manage, SLA adherence is 100%.
What I took from it
Day-one performance is decided weeks before day one. The best way to protect a launch is to settle readiness questions (capacity, training, access, cover) while they are still cheap to fix.