Near-realtime discovery
Instead of delivering a 40-page assessment, you build an MVP the client touches by the second call.
Open roleHUMAN FACTORS TECHNOLOGY
Forward Deployed AI Builder.
Someone who listens to the problem and while listening opens their editor. Not a wireframe, real code that runs. When the client changes their mind, you tweak three lines and show the new version two minutes later. A consultant who thinks by generating, and teaches while doing.
01 /Who we're looking for
Thinks by generating. Teaches while doing.
Sits with the client, listens, and opens the editor.
Picture the scene: a room with an area head, a CFO or operations lead. You listen to the problem. While you listen, you open Cursor or Claude Code and start building the solution. When the client says 'but what if I wanted also...', you tweak three lines and show the new version two minutes later.
At the end of the session, you turn the screen toward the client's IT team and explain what you did, line by line. Then you turn it toward the business and say 'you could have done this too, here's how'.
It's a new role. It's rare. It's hard to find. It's exactly what we're looking for.
02 /What you'll do
Six things, every week.
Live iteration
You design the solution as the business tells it. You iterate in real time, you don't postpone to 'I'll send a proposal Monday'.
Working prototypes
Cursor, Claude Code, Lovable, v0, Bolt, n8n: whatever moves thinking fastest toward the working thing.
Decide when to pass it
When the MVP is good enough, when it goes to the AI Solution Engineer to be hardened, when it's to be thrown out because we're solving the wrong problem.
Live pair programming
With the client's tech team: you don't replace them, you contaminate them. By end of day they know things they didn't.
Show business what they can do
With today's AI tools, on their own. You lower the bar, you don't keep it. The value is transfer, not monopoly.
03 /Expected stack
Vibe-fluency, not certifications.
Daily tools
- Cursor, Claude Code, Codex CLI: used daily, like an editor
- Native prompt engineering: you talk to a model like to a colleague
- Python and TypeScript at a conversational level (no PhD, fluency needed)
- Prototyping: Lovable, v0, Bolt, Streamlit, minimal Next.js
Fundamentals
- Anthropic and OpenAI LLM APIs at least. You understand differences, costs, latencies, context window
- Database and API: decent SQL, REST by memory, webhooks without panic
- Serious bonus: you've already published something of your own (repo, side project, agent, tool, even small)
You don't need the boss's permission to build. When you see an idea, you ship it.
04 /Required mindset
This matters more than the stack.
Curiosity first
When a new model drops, you try it that same day. When you spot a weird workflow, you want to know why.
No fake enthusiasm
AI generation still excites you. If you're already jaded, you're not the right fit. We want vibrant people, not tired ones.
Bilingual tech-business
You lower the register with a sales director, raise it with a solution architect, without changing the substance.
Not protective of your code
What you build goes to the client or to a colleague without drama. You document, you pair, you pass the ball.
At ease with uncertainty
The first MVP is bad. The second is embarrassing. The third works. You know it's normal, and you iterate without panic.
Natural teacher
When you explain, people feel more capable, not smaller. This, above all.
05 /What this role is NOT
Four profiles we're not looking for.
The AI engineer who waits for specs
Doesn't ask for written specs before starting. Builds as the problem becomes clear.
The slide consultant
Doesn't talk AI without having ever touched a repo. No repo, no conversation.
The hackathon demo-er
Doesn't make pretty but unfinished things. Things that break at the client's third question don't count.
The lone wolf
The value here is capability transfer, not personal output. No prima donna.
06 /Where you can come from
Four typical paths. Yours might be different.
Full-stack developer
You fell in love with AI in the last 2 years. The previous framework bores you, AI excites you.
Junior or mid consultant
You went technical, now you build. Slides weigh you down, the editor relaxes you.
Founder of a small thing
It didn't scale. But you learned a lot. That kind of lesson is taught by no one.
Tired data scientist
Of notebooks. You want to ship. You want to see the client use what you built, not just a report.
Experience required: 3-6 years. Mid-seniority is perfect. If your path differs from these four but you recognise the mindset, apply anyway.
07 /How we work
Time, autonomy, contract form. Seriously.
Short week
The Hestro working time is 4 days out of 7. It's not a pilot, it's the standard.
4/7Remote-first Italy
You decide where to work from. Travel 1-2 days a week to clients.
Flexible contract
Employee, consultancy, ongoing VAT, hybrid. We pick together.
Side projects welcome
Parallel activities (advisory, teaching, open source) are an asset.
08 /How we select
Three steps. Step 2 is a live build.
Intro call
With one of the Hestro seniors. We want to know where you come from and what excites you today.
Live build session
We open the editor together. Anonymised real case. We build, iterate, think out loud.
Shadowing day
Stay with us on a live project. Understand if you recognise yourself in how we work.
HUMAN FACTORS TECHNOLOGY
Ready to apply as Forward Deployed AI Builder?
Go to the contact form on the Careers page. Tell us 'Forward Deployed AI Builder' in your message, attach your CV or link your professional profile, and tell us in two lines what you're building right now.
Reply within 5 business days