Brahim Bousnguar

Client work

API-led integration for Roche Bobois

Replacing point-to-point flows with one observable integration backbone on MuleSoft

The retailer's e-commerce site, ERP, CRM and a handful of third-party services were talking to each other directly. Every new need meant another point-to-point flow, and when one failed it was hard to tell where.

Since January 2025 my job has been to move that onto MuleSoft Anypoint, API-led: wrap each system once, and let every flow reuse it.

The shape

Channels

E-commerce

  • Orders
  • Customers
  • Catalogue

Integration layer

MuleSoft Anypoint

  • Experience APIs
  • Process APIs · orchestration
  • System APIs · DataWeave

Back office

ERP · CRM · partners

  • ERP
  • Salesforce CRM
  • Third-party services

System APIs own the connection to each back-office system. Process APIs orchestrate the business flows. Experience APIs serve the channels. Change one system and only its System API has to move.

In numbers

  • 9 Mule applications, and I'm the only Mule developer on them.
  • 3 service integrations built end to end by me: quotes, products and cases.

What I built

  • Reusable APIs following API-led connectivity. I moved development from Anypoint Studio (Eclipse) to Anypoint Code Builder (VS Code).
  • The quote, product and case integrations, from the first flow to production.
  • A stable customer master-data flow: I took over the single-customer-repository integration and stabilised it.
  • DataWeave transformations for mapping, enrichment and orchestration, plus a shared DataWeave library for the mappings.
  • A shared error-handling library with Jira integration.
  • CI/CD for deploying the Mule applications.
  • Integration flows between the e-commerce platform, the ERP, Salesforce CRM and third-party services.
  • API specifications, best practices and governance standards. I also sit in the design-authority meetings.

Where it landed

One observable integration backbone across the order, CRM and ERP flows, instead of a web of direct links. The governance standards became the team's way of working.

What came out of it

AI is part of the daily workflow on this mission: GitHub Copilot, an Ollama-based CLI that writes commits and PRs, Claude and Codex for code review.

It also produced a side project. I first built a log-reading MCP server for this estate, then released a client-free version as mulewatch.

Working on something similar? Say hello.

Plain text: api-led-luxury-retail.md · All client work

Contact

Say hello.

b.bousnguar@gmail.com

SAP Commerce Cloud, MuleSoft and Salesforce integration · Nantes · EN / FR