TVH Parts.

The Situation

TVH Parts operates one of the largest B2B e-commerce platforms in the industrial parts sector, processing tens of thousands of orders daily and driving 1.2 billion in annual revenue. The frontend had grown organically over years within a large NX monorepo, accumulating technical debt that was becoming a barrier to growth.

The problems ran deeper than code. Frontend engineers were fractured across Run and Build teams with no unified technical direction. A Center of Excellence existed in name, but lacked the authority or resources to enforce standards. Engineers embedded in other teams received conflicting technical guidance, had limited career growth, and operated in silos.

The codebase reflected this fragmentation. Code reviews were handled by junior developers with no senior oversight, focusing on functional correctness rather than performance or maintainability. Core React principles were poorly understood across the team, leading to unoptimized rendering, inefficient state management, and unnecessary complexity. Redux was being misused in ways that negated its benefits and inflated the codebase with boilerplate.

The component library lived in a separate repository with a painful publish cycle, was owned by the wrong team, and had grown wildly inconsistent. Designers bypassed the system entirely. Multiple teams were independently solving the same problems, resulting in duplicate implementations across the platform.

There was no frontend observability. The existing monitoring stack covered backend and infrastructure health, but provided zero visibility into frontend performance, rendering issues, memory leaks, or error tracking. Problems were discovered reactively through user reports, not proactively through tooling.

Developer hardware was inadequate. Engineers were offloading basic tasks like linting and testing to CI/CD pipelines because their machines couldn’t handle them locally, creating pipeline congestion and wasting time across the team.

Agile practices had deteriorated to Scrum in name only. The product backlog had grown unwieldy and opaque, sprint planning lacked structure, and retrospectives weren’t yielding improvements. Metrics were exported from Jira manually, producing stale and inaccurate data.

On top of all this, a legal deadline for accessibility compliance was approaching with no clear path to remediation.

What We Did

We embedded as fractional frontend leadership, taking ownership of the technical direction across the entire frontend engineering organization.

Authored the modernization roadmap. We produced a comprehensive executive summary and detailed roadmap covering every dimension of the problem: codebase quality, architecture, team structure, tooling, observability, accessibility, and process. This wasn’t a PDF to collect dust. It was the document that secured stakeholder buy-in and budget for the entire initiative.

Empowered the Center of Excellence. The COE needed authority, not just a title. We defined its expanded mandate: ownership of coding standards, architectural governance, knowledge sharing, platform improvements, resource allocation, and mentorship. We made the case to leadership and got it approved.

Restructured engineering teams. We addressed the Run/Build split and the problem of embedded engineers operating without clear technical oversight. We proposed and evaluated team structures, from integrated feature teams to models that could support future Fusion Teams, ensuring all engineers had consistent technical direction and career development paths through the COE.

Defined the technical migration strategy. We authored Architecture Decision Records for every major technical shift: migrating from Redux to TanStack Query for state management, adopting TanStack Router for type-safe routing, transitioning to Tailwind CSS with shadcn/ui patterns, evaluating Next.js for SEO-critical pages, and establishing API-first development workflows with OpenAPI contracts.

Took control of the component library. We moved it into the monorepo, transferred ownership to frontend engineering within the COE, and defined a governance process that prevents future fragmentation. We enforced design system adherence and initiated a full accessibility audit.

Introduced frontend observability. We evaluated and proposed a dedicated frontend monitoring solution to replace the reactive status quo. This included error tracking with meaningful stack traces, performance profiling, automated alerting, Jira integration, and session replay capabilities. We designed the pilot program and evaluation criteria.

Raised the bar on code quality. We implemented structured code reviews with mandatory senior engineer oversight, developed and enforced coding standards, established automated linting, and pushed for investment in developer training through platforms like Frontend Masters.

Fixed developer experience. We made the case for hardware upgrades to high-performance machines, eliminating the absurd cycle of developers waiting on CI pipelines for basic linting feedback. We optimized local development workflows to remove daily friction.

Addressed process gaps. We worked with the Agile coach to implement structured practices: backlog refinement, focused sprint planning, consistent ceremonies, meaningful retrospectives, and automated metrics collection to replace the manual Jira exports.

The Result

The modernization is ongoing, but the foundation is solid. The COE now operates with real authority. Engineering teams are unified under consistent technical direction. The architectural roadmap is approved, funded, and being executed. Migration work on TanStack Query is actively in progress.

The patterns, processes, and governance structures we established were designed to outlast the engagement. The team owns and continues to execute the roadmap independently. That was always the goal: leave them in a position where they don’t need us anymore.