W062017–19DeliveryEcommerce, global24+ engineers
Outcome
35%
faster sprint velocity after standardizing ceremonies and metrics across a distributed team of 24+, with planning time cut in half.
11releases shipped, including 3 hot-fixes, with every pre-existing bug closed out
50percent reduction in sprint planning time after standardization
10million-dollar program budget delivered across 85,000+ billable hours

Upgrading the platform under a peak-season deadline that couldn’t slip

01 / Context

Large-scale ecommerce platform on SAP Commerce Cloud (Hybris), managing distributed engineering teams of 24+ across 6 time zones, including 3 direct reports in PM roles.

$10M+ in program budget, 85,000+ billable hours, coordinating platform integrations that required hardware/software alignment across distributed manufacturing and fulfillment networks. Global team: SF (PM/BA), NYC (principal tech lead), Guadalajara (DevOps/dev/QA), Chișinău (dev), Buenos Aires (dev).

02 / Problem

Three things had to land at once, and none of them could slip: a platform upgrade, a new country launch, and a backlog of bugs that predated the project.

Hybris 5.7 needed to become 6.6, an Australian storefront needed to launch from scratch, and existing bugs needed remediation — all before peak holiday traffic, non-negotiable.

INCREMENTAL HYBRIS UPGRADE — DIAGRAM
FIG 01 — 5.7 to 6.6, incrementally, not in one leap.

03 / Approach

Take the safer technical path, then run delivery and program management as one job, not two.

01Upgrade Hybris incrementally rather than in a single version leap — a technical decision made early that shaped everything downstream.
02Run the program as ScrumMaster and Program Manager simultaneously: team charter, sprint planning, backlog grooming, demos, and retrospectives on the delivery side; change control, RAID/risk register, RACI, release and launch planning on the program side.
03Fold QA into dev estimation from the start once the incremental-upgrade approach made constant regression testing unavoidable, and have QA write automation scripts as part of CI.

04 / What I did

Orchestrated the international launch (the client’s new Australian storefront domain) on time and on budget despite timezone complexity. Implemented standardized sprint ceremonies and metrics tracking across the distributed team, improving velocity 35% and cutting sprint planning time in half.

Managed the full tech stack coordination — Nexus, Jenkins, Gitflow, Bitbucket, Apache Solr, JMeter for peak-ready performance testing, Groovy/Spock for unit testing, Java backend, Avalara for tax, RabbitMQ, Braintree, PagerDuty, Chef, Angular.js frontend — while also running budget/resource planning against PTO and regional holidays across 5 countries.

Role: ScrumMaster & Program Manager · Digital commerce consultancy · May 2017–Sep 2019 · Team of 24+ across 6 time zones

05 / Outcome

11 releases (3 hot-fixes), every pre-existing bug fixed, the Australian site launched in time for the holiday season, and every project landed within budget because scope changes were managed and communicated with the client throughout. The Hybris upgrade itself launched in February 2019 — after peak season — because of delays in client-side scope sign-off, not delivery.

SPRINT VELOCITY, 24+ ENGINEERS — DATA VISUAL
FIG 02 — Velocity up 35 percent after standardizing ceremonies.

06 / What I would do differently

I’d have required peer review on major technical decisions from day one. The incremental-upgrade approach was the right call, but we made it without that check, and constant regression testing became the tax we paid for skipping it — trust, but verify.

I’d have flagged my own bandwidth sooner. I was working 10–12 hour days for months because we couldn’t hire or contract additional PMs — that’s a resourcing risk I should have escalated earlier instead of absorbing it.

I’d have built the ‘Sprint Mood’ survey on day one, not after noticing engineers were going quiet. A cheap, anonymous channel for team temperature should be standard, not a reaction.