The AI engineering role you will not have to hire for.

An AI engineer deployed directly inside the client's team: he understands the real problem, writes production code, connects AI to the existing systems and stays accountable until the solution runs. The model comes from Palantir and is now adopted by OpenAI, Anthropic and Google. At AI Makers it is the core of the offer: an FDE embedded in your team, supervised by our CTO.

Because 95% of enterprise AI pilots produce no measurable return (MIT study, 2025): the models are powerful, but wiring them into a real environment is the actual problem. As a result, FDE job postings multiplied by 8 in a year, and the AI labs pay this profile between $385K and $1M a year. Few companies can hire that profile. Many can deploy one.

You do, entirely. Code, playbooks, documentation: all the intellectual property is yours, and it is written into the contract. The day we leave, everything stays with you. Zero dependency, zero hostage-taking.

Walid, our CTO. Every system delivered is reviewed with him, every technical trade-off goes up to him, and the engineer draws on the playbooks of the 200+ systems the studio has already deployed. Your engineer works inside your team, never alone with a problem.

The goal fits in one word: autonomy. At 6 months the systems run, the playbooks are written and your leads are trained to keep them alive. Every system leaves with its documentation, in full client ownership.

Both. We work on site in mainland France and in Morocco, from our Paris and Casablanca offices, and remotely across the whole French-speaking region. The format is set around your organisation at the diagnostic.

3 months. The engagement then continues monthly, with 30 days' notice. The first weeks go to the audit and the costed roadmap, then the first systems ship to production from the first month.

You can, and some of our clients do afterwards. The question is timing: hiring a senior AI engineer takes 6 to 12 months and commits €101,000 to €122,000 fully loaded per year, for a role whose shape you do not yet know. Deploying an FDE takes a few weeks, costs a fraction of the role, and teaches you exactly what that role must do at your company. You then hire with your eyes open, or not at all.

That is expected, and it is even a good exit scenario. The systems, the playbooks and the documentation belong to you from day one. The day you hire your own engineer, they inherit a documented environment and teams already trained: they are productive within weeks. We can also help you define the job spec and assess candidates.

A freelancer sells you days and leaves with the know-how. We sell an outcome. When we leave, your leads stay, the documentation stays, and all the code belongs to you.

Python, TypeScript and SQL on languages, modelling and ETL on data, APIs and cloud on architecture, RAG and multi-agent systems on applied AI. The half that really decides is elsewhere: requirements gathering, trading scope against speed, clear writing, stakeholder management. Those four field skills are acquired by delivering at clients.

Between €45,000 and €55,000 gross for a junior in the Paris region, €55,000 to €75,000 for a mid-level, €75,000 to €100,000 for a senior, and above €100,000 for a lead. Add around 35% for the fully loaded cost. The regions sit about 15% below. In the American labs the same profile negotiates between $385,000 and $1 million.

The solutions engineer works in pre-sales and demonstrates a product. The Forward Deployed Engineer arrives after signature and builds inside the client's environment. The first carries a sales cycle, the second carries a system in production.

On applied AI, often yes. An IT services firm staffs a profile on a scope defined in advance, with a best-efforts obligation. The deployed engineer commits to an outcome measured against a KPI, system by system, and trains your teams at the same time.

Zero days, because onboarding happens before. Two weeks before kick-off he reads your processes, learns your vocabulary and gets to grips with your tools. The first workstream opens in week 1.

According to the MIT study published in 2025, the failure comes from the last mile of deployment rather than from the quality of the models. Access rights, edge cases, real adoption by the teams, integration with existing systems. Every enterprise environment is complicated in its own way, and no off-the-shelf software plugs into it alone.

It depends on three things: the domain depth required, how stable your workload is over three years, and what you already know about the role. If the critical knowledge takes two years to acquire, or if your workload is stable, hire. If you are opening this role for the first time, deploy first and hire afterwards knowing what to look for.

Three cases. Extreme domain depth, where a permanent hire will carry the knowledge better. A stable, predictable long-term workload, where an employee costs less. A team already in place, where what is missing is method rather than hands — in which case we point you to our corporate training.

We change engineer, within a week. The successor picks up written playbooks and a documented environment. That is a fundamental difference from a failed hire, which costs six more months.

That is the standard case. He plugs into your security processes, your code reviews and your environments. All the code goes to your GitHub, versioned and reviewed. Our AI governance page details the framework.

It is already included in the engagement: 2 hours a week, on the systems your engineer is building at your company. Your developers scope and ship paired with him, code review goes through our CTO, and the playbooks they write become your reference. If you are looking to build skills without a deployment, our corporate AI training covers the foundations.

On the scoping, trade-off and stakeholder side, yes, and it is often the fastest profile to bring up to speed. What they lack is the technical foundation. Our best pairs put a senior developer and a product manager on the same workstream.