Case Study · Fullstack product

REST analytics: API + web app

Problem: Ad hoc exports and mismatched API/dashboard models slowed customer analytics workflows.
Role: Modeled analytics in Prisma, versioned Express APIs, and aligned React/Vite dashboards to API-enforced RBAC.
Stack: TypeScript, Express, Prisma, React, Vite
Impact: 18 versioned API routes · KPI query P95 5.1 ms → 2.4 ms (benchmark)
Links: GitHub · Live (demo tenant)

A customer analytics product where typed API contracts and RBAC match what the React dashboard shows—one source of truth from route handlers to segment views, without duplicating business rules in the UI.

Product guarantees: analysts see the same aggregates the API enforces, admins get explicit lifecycle controls, and new features ship through shared schemas—not parallel “backend” and “frontend” interpretations.

Context: baseline came from pre-layered API patterns and manual role handling; post-change measured across shared QA and contributor onboarding cycles.

Product + API

Customer analytics with Express + React + Prisma

Context: Teams needed lifecycle CRUD on customers alongside KPI and segment views—without duplicating business rules in the UI.

Problem: Customer operations and analytics lived in separate flows, creating policy drift, duplicated validation, and slow onboarding for new contributors.

What I built: Implemented a layered Express + TypeScript API (routes → controllers → services), JWT access plus refresh in HttpOnly cookies, Prisma migrations as schema source of truth, and a React + Vite dashboard consuming typed API contracts for customer CRUD and aggregate analytics endpoints.

Impact: Unified product and analytics workflows into one contract-first stack with clearer ownership boundaries, less rule duplication, and faster iteration on API + UI changes.

Tradeoff I made: Chose React + Vite and Express for speed and lower ceremony in a focused dashboard product; preserved a migration path to Next/Nest patterns as team size and surface area grow.

Proof artifact · route contract

Analytics endpoints enforced at the API layer

GET  /api/customers/:id/summary   → analyst, admin
POST /api/customers                 → admin
GET  /api/analytics/segments        → analyst, admin
PATCH /api/customers/:id/lifecycle  → admin

// Shared Zod schemas exported to React client
export type SegmentSummary = z.infer<typeof segmentSchema>;

Product chrome—the only case with a dashboard surface as a first-class artifact.

Layered

REST: routes → controllers → services → data

Baseline: 5 ad-hoc endpoint patterns
After: 1 layered convention
Sample: Shared QA + contributor onboarding

RBAC

admin and analyst roles with explicit boundaries

Baseline: Implicit UI-only role checks
After: API-enforced role guards
Sample: 18 API routes + /health

4 days

median time to ship a scoped feature

Baseline: 11 days (hand-rolled API + UI)
After: 4 days with shared contracts
Sample: 8 comparable features

Architecture diagram

React dashboard -> Express API layers -> Service and business rules -> Prisma/MySQL model -> Analytics aggregates

Before & After

Before

Analytics and operational edits lived in different places, and API boundaries were unclear to new contributors.

duplicated rules risk + higher integration cost

After

One coherent stack: typed contracts from DB to UI, explicit auth, and a single place to evolve aggregates and access policy.

clear ownership: API vs dashboard vs data layer

Execution Footprint

Stack aligned to the REST-app reference project in this portfolio workspace.

Tech stack

Node.js

TypeScript

Express

Prisma

MySQL

React

Vite

Techniques / security / data

JWT + refresh in HttpOnly cookie Layered REST handlers Migrations as source of truth SQL for aggregate analytics

REST analytics: API + web app

Product + API

Customer analytics with Express + React + Prisma

Architecture diagram

Before & After

Execution Footprint

More case studies

Data Platforms

Agent Benchmark Engineering

Inference & Evals

Fullstack product

Backend & Cloud

SENTINEL