The situation
In wealth management, downtime isn't an inconvenience — it's clients locked out of their money and advisors locked out of their book. This migration had to move an enterprise platform underneath more than two million users with no service interruption, coordinated across more than eight vendor teams, none of whom reported to each other and all of whom had their own release schedules. The failure mode of programs like this is well known: dependencies discovered at the seam between vendors, in production, at the worst possible moment.
What I did
The work was dependency choreography — turning eight-plus independent vendor release schedules into one release train with a single view of sequence, risk, and rollback. Every cutover step had a rehearsed rollback protocol; every dependency at a vendor seam had a named owner on both sides.
- Orchestrated release-train dependencies across 8+ vendor teams — one integrated sequence instead of eight independent schedules, with the seams between vendors treated as the primary risk surface.
- Built and rehearsed rollback protocols for the cutover path, so every step forward had a tested step back — the discipline that makes "zero downtime" a plan rather than a hope.
- Held delivery accountability across organizations that didn't report to each other, through escalation paths and reporting cadence that vendors respected because they were used.
The outcome
The migration completed with zero downtime for 2M+ end users — ahead of schedule and under the resource plan. The client's confidence showed up the way it always does in financial services: $10M+ in follow-on work awarded on the strength of the delivery.