Hi, I am

Wassim Mhiri

I am a

I build the systems that let AI agents ship production code safely. The answer turned out to be process, not prompting.

Sfax, Tunisia

WM

About

Seven years building enterprise software — architecture, APIs, and the modernisation of large systems on .NET, Angular and Azure. The last stretch went into a harder question: how do you hand work to an AI agent without losing control of the system it is changing?

0
years in enterprise software
0
developers led as team lead
AZ-204 / AZ-400
Microsoft certified

Most teams have AI autocomplete. Very few have a way to let agents ship.

Tab-completion is easy to adopt and easy to contain. Agents are neither. The moment an agent can edit many files, open pull requests and touch infrastructure, the bottleneck stops being the model and becomes everything around it — whether the intent was written down, whether the architecture is legible, whether anyone can tell what changed and why.

That is an engineering-process problem. I test every answer on my own platform before it goes anywhere near a client's codebase.

  1. Intent
  2. Spec
  3. Plan
  4. Build
  5. Review
  6. Ship
01

An agent that writes code does not approve it

The thing that produces a change and the thing that reviews it must be separate, with a human holding the merge. Self-review by the same context is not review.

02

No spec, no code

Every change starts as written intent, becomes a specification, then a plan, and only then code. The questions nobody answered surface on day one, when changing the answer is still free.

03

Architecture decisions get recorded

Decisions live as numbered records. An agent that can read why a rule exists stops arguing with it, and so does the next engineer.

Skills

Where the seven years are, the practice I bring to a team, and what I have shipped to production with agents alongside me.

Core expertise

Where the depth is.

  • .NET (Framework & Core)
  • C#
  • Entity Framework
  • Angular
  • TypeScript
  • RxJS
  • Microsoft SQL Server
  • T-SQL
  • Microsoft Azure
  • Azure DevOps
  • Microservices
  • Clean Architecture
  • API development
  • Technical leadership
  • Agile / Scrum

AI-driven development

The practice itself.

  • Agentic SDLC
  • AI agent orchestration
  • Model Context Protocol (MCP)
  • Spec-driven delivery
  • Prompt engineering
  • Claude Code
  • Copilot

Delivered through agentic workflows

Shipped to production with agents, under spec-driven review.

  • React
  • Node.js
  • Fastify
  • PostgreSQL
  • MySQL
  • AWS
  • OpenTofu / Terraform
  • Docker

Experience

Seven years, mostly on platforms people depend on.

Senior Software Engineer · ClearCo

Feb 2026 — present

Enterprise HR platform, onboarding for the US market.

I own technical analysis and architecture decisions for the onboarding domain, and run two fronts at once: building the new platform while keeping the application currently in production healthy. I also introduced the team's agent-assisted development workflow and the guardrails that make it safe to merge — specifications written before code, architecture decisions recorded as they are made, and a clear separation between the agent that writes code and the review that approves it.

  • DocuSeal — electronic signature built into onboarding, so new hires sign their documents digitally instead of on paper
  • Backbone.js and ASPX modules progressively migrated to React and TypeScript
  • Services on AWS with infrastructure provisioned as code in OpenTofu
  • React
  • TypeScript
  • Fastify
  • .NET
  • PostgreSQL
  • AWS
  • OpenTofu
  • AI agents
  • MCP

Founder & IT Consultant · Independent

Feb 2026 — present

Consulting on agentic delivery and enterprise platform work.

IT consulting and software engineering for enterprise clients — scalable cloud solutions, web applications and modernisation strategies. I also run my own multi-repository platform as a lab for agentic delivery: specifications before code, architecture recorded as decisions, agents operating under enforced constraints. What survives real use goes to clients; what does not gets thrown away.

  • Agentic SDLC
  • Architecture
  • .NET
  • Angular
  • Azure DevOps

IT Team Lead · EUREXO, part of CED

Jun 2024 — Jan 2026

Prospect Eurexo — claims management, localised for the French market.

Led ten developers building the French evolution of the Prospect platform: configurable product, client and contract management, localised workflows, and case management built to scale. I spearheaded the architecture decisions, managed sprint priorities and resource allocation, and used Azure tooling to keep API and database health visible.

  • Quintus — a Power App used for calculation and compensation management
  • Sinapps DARVA — property damage claims, improving workflow between stakeholders
  • Espace Assuré — self-service portal for scheduling visits, tracking dossiers and uploading documents
  • IFS Planning Tool — optimised planning and resource allocation
  • Angular
  • .NET
  • Microservices
  • Azure
  • Team leadership

IT Team Lead · CED

Jun 2023 — Jan 2026

Prospect Platform — the configurable claims platform underneath it all.

Managed an eight-person engineering team building a highly configurable claim management platform. Architected scalable workflows for dossier creation and contract management, integrated external applications, and enforced standards through code review, automated testing and performance profiling. Mentored junior and mid-level developers.

  • .NET
  • Angular
  • Azure
  • Code review
  • Mentoring

Projects

The platform I use as a lab for agentic delivery. Names and client details are withheld.

Admin dashboard listing venues

2026 — ongoing

Online Booking Platform

A multi-tenant reservation platform for service businesses: a web admin in Angular, a mobile app in Flutter and a .NET API on EF Core. Built as the lab for my agentic delivery practice — specifications before code, architecture recorded as decisions, and agents working under enforced review.

  • Angular
  • Flutter
  • .NET
  • EF Core
  • Agentic SDLC

Education

Engineering training, plus the certifications that keep the Azure side current.

Education

  • National School of Computer Science (ENSI)

    Engineer's Degree in Computer Science

    2016 — 2019

  • Sfax Preparatory Engineering Institute (IPEIS)

    Preparatory Engineering Diploma

    2014 — 2016

Languages

  • Arabic · Native
  • French · Fluent
  • English · Fluent

Certifications & training

  • Microsoft Certified: Azure Developer Associate (AZ-204)

    Microsoft

  • Microsoft Certified: DevOps Engineer Expert (AZ-400)

    Microsoft

  • Project Management Professional (PMP) — training

    Sep 2025

  • Advanced .NET

    Advancia · Sep 2023

  • Advanced SQL

    RFC · May 2023

Contact

If you are working on any of this, I would like to hear about it.

Agentic engineering practice, AI-assisted delivery, or enterprise platform work — consulting engagements or just a conversation.