Back to work
/ 05Platform Engineering

Storefront API Hardening & Digital Inventory Platform

Hardened a high-traffic storefront API, delivered taxonomy-driven breadcrumbs against a fixed release window, and turned a one-off pooling audit into a reusable engineering practice adopted across the client's pod structure.

01

Breadcrumb feature shipped on time

Unblocked the CMS roadmap and improved the shopper browse experience.

02

Weekend fire-drills eliminated

Post-release iOS error spikes turned into a tracked, declining trend.

03

Audit became a practice

Pooling methodology adopted across pods — compounding uptime gains beyond the original scope.

04

500+ legacy lines retired

No customer-facing changes — reduced maintenance burden on a high-traffic service.

05

Zero inventory-sync failures

Held through outage conditions, protecting stock accuracy and preventing oversell.

The Challenge

The storefront API — a C# / .NET / GraphQL service ingesting from Kafka and reading across SQL Server and Redis — had grown organically across releases. Unreachable code paths in a dedicated caching service gave the illusion of error handling that never actually ran. Connection pooling behavior (HttpClient and SQL) was inconsistent across the API surface, but no one on the in-house team had a clean read on it. Release weekends were periodically generating tens of thousands of post-release iOS bad-request spikes that pulled the platform team into scramble responses.

At the same time, the product roadmap required a new breadcrumb feature for the CMS UI that depended on ingesting taxonomy from Kafka and reconciling state across SQL and Redis under tight release timing.

Our Approach

We embedded three senior engineers into the client's pods over a multi-quarter engagement.

A senior backend engineer led the breadcrumb / taxonomy ingestion feature end-to-end: designing a Kafka consumer, writing the new taxonomy table via migration, and reconciling changing values into Redis. The same engineer simplified the resulting GraphQL resolver work, retiring more than 500 lines of legacy code without changing consumer behavior.

A second engineer took ownership of a connection-pooling audit across HttpClient and SQL, ran load-testing sessions with the platform team, documented findings, and presented the methodology in a cross-team demo — turning the audit into a reusable engineering practice that other pods adopted for their own services.

A senior QA-focused engineer led testing strategy for Subscriptions and Payments initiatives across web, iOS, Android, and AEM, plus PI-level planning and cross-team alignment with partner vendors. The same engineer helped stand up automated regression validation with SauceLabs for release weekends.

Engagement Model

Three embedded senior engineers, full-time, multi-quarter. Pod-aligned reporting into the client's release-management cadence.

Technology
C# / .NETGraphQLSQL ServerRedisKafkaAzure DevOpsGitHubSauceLabsWSL2

Ready to get back to building?

Tell us about the engagement. We typically respond within one business day with a named builder who can talk substance — not a generic sales pitch.

Prefer email? info@inthe.cloud