Rebuilding a loyalty platform for thousands of e-commerce merchants

Real project, from my time as Senior Developer & Team Lead. The scope figures below are accurate; the bracketed spots are where you can drop in the specific performance numbers you’re comfortable sharing publicly.

The context

A loyalty platform serving thousands of e-commerce merchants had outgrown its original design. The core was a .NET monolith: dependable, but increasingly expensive to change, hard to scale selectively, and risky to deploy. Every new feature touched everything.

What I led

• A monolith-to-microservices rebuild. I led the initiative to re-architect the platform into a modern, service-based system — splitting the loyalty domain (points, rules, tiers, activity) into services that could ship and scale on their own — alongside a new React dashboard for merchants.
• A large-scale migration. I orchestrated moving thousands of merchants and millions of records from the legacy system to the new platform, including the careful transformation of points-activity history and loyalty rules into the new data model — the part where mistakes are visible to real customers.
• The integration surface. The platform connected to Shopify, BigCommerce, and custom systems, plus marketing tools like Klaviyo and Attentive, so merchants’ loyalty programs worked wherever they sold and messaged.

How it was done

Migrations like this fail quietly, so the discipline mattered: explicit contracts for the legacy and target shapes, batch-by-batch transforms with reconciliation at each step, and the old system kept authoritative until the new one earned trust. I worked closely with cross-functional teams and leadership to surface blockers early and keep progress transparent.

The outcome

The payoff showed up where merchants and shoppers actually feel it: loyalty widgets and dashboards that used to load slowly now come up at industry-standard speeds, in line with what shoppers expect from a storefront. Underneath that, the move from monolith to services brought real gains in scalability and reliability — and left a platform the team could evolve feature-by-feature instead of all-at-once. The clearest sign was how the work itself shifted: before the rebuild, roughly 80% of our effort went to bug fixes; after, close to 90% went to shipping new features — the team moved from firefighting to building.