// Case studies
Engagements, anonymised
The kind of production problems teams actually bring me. Client names and identifying details are stripped out, but the architecture, the failure modes and the fixes are real.
A look at the kind of problems businesses actually bring me. Client names and identifying details are left out, but the situations and the work are real.
Giving a fast-growing product's backend real headroomGetting a growing product ready for serious growth
A consumer product's order processing lagged under peak load and a key third-party integration failed silently, on infrastructure built from quick fixes.A company had found genuine demand for its product, but the infrastructure behind it was a patchwork of quick fixes that was just about coping.
Read the case study →Running a tech debt remediation programme across a multi-service platform, without a feature freezeLeading a technical debt clean-up without stopping the business
Years of accumulated shortcuts across a multi-service platform meant slow delivery, regressions across service boundaries, and releases nobody fully trusted.Years of shortcuts across a large, multi-service platform had made every new feature slower to build and every release riskier than it should be.
Read the case study →Taking a founder-only codebase to a real engineering teamBuilding the first real engineering team around a founder
A B2B product had a bus factor of one. The founder was the only engineer who understood the system, and every change, down to small bug fixes, went through him.A founder had built the whole product himself and every technical decision, big or small, still had to go through him personally.
Read the case study →Hardening a 20M-visit platform for traffic spikes and calm deploysMaking a 20 million visit platform boring again
A content platform serving 20M+ visits a month degraded under traffic spikes, and releases were manual, out-of-hours and hard to roll back.A content platform with more than 20 million visits a month was buckling under its own traffic spikes, and every deployment felt like a gamble.
Read the case study →A shared core and a per-franchise rules layer for a multi-tenant freight platformGiving a multi-franchise shipping platform an architecture it could grow into
A multi-tenant freight platform was absorbing per-franchise pricing and workflow differences as special cases in shared code, so every new franchise risked regressions for the existing ones.A shipping platform serving multiple franchises and their customers needed an architecture that could handle different rules for each franchise without collapsing into chaos.
Read the case study →