Lower cloud cost. Keep the storefronts standing.

TTAC ran a high-volume digital publishing and ecommerce operation across three properties, with WordPress, WooCommerce, Shopify, and Laravel systems feeding fulfillment, customer service, marketing, and revenue. The infrastructure had grown the way most do, under deadline, in layers, with each decision locally reasonable and the aggregate expensive and fragile.

The constraint that shaped everything: launches. Documentary releases repeatedly drew more than a million viewers in compressed windows. Cost reduction was necessary, but a cheaper platform that fell over on launch day would have been a bad trade. The work had to move economics and reliability in the same direction.

I reported to executive leadership as the primary technical point of accountability and led a distributed engineering team, while remaining personally responsible for architecture decisions and hands-on delivery. The role required both management and hands-on work.

Recurring AWS spend moved from roughly $15,000–$20,000 per month to about $1,000. That did not come from a single dramatic change. It came from treating the bill as an engineering problem: architecture modernization, governance over what was provisioned and why, Spot capacity where workloads could tolerate interruption, migration onto managed services where operations overhead exceeded license cost, and a CloudFront redesign that changed what the origin had to serve.

Each layer reduced spend; together they reduced it by more than 90%. The governance piece mattered as much as the technical piece; cloud cost is usually a behavior problem before it is an architecture problem, and the controls had to outlive any one migration.

Commerce operations had grown into four disconnected surfaces, WordPress, WooCommerce, Shopify, and Laravel, each with its own view of orders, fulfillment, customers, and analytics. I unified them through an event-driven AWS pipeline, so operational events had one path and one meaning instead of four integrations that disagreed with each other.

On top of that foundation I launched three Shopify Hydrogen / Next.js storefronts with subscription analytics, and migrated delivery infrastructure from Puppet-managed configuration to GitHub Actions. Release cadence roughly tripled. CI/CD work pays off by removing the coordination cost of shipping.

The accumulated results: more than 90% lower recurring cloud spend, 99.97% uptime across three ecommerce properties, approximately threefold release cadence, and repeated launches sustaining over a million viewers. These outcomes compound each other: the cheaper platform was also the more reliable and faster-shipping one, because the same architecture and governance work drove all three.

These are team-era outcomes under my technical direction and personal implementation. The figures describe the systems' performance; they are not a claim that any single change independently caused every result.