Project Lead

On-site in Barcelona, on-site·Added today

Sovrano AI

5 open roles

Apply

Overview

Requirements

Have the right to work in Spain
Sovrano AI doesn't mention sponsorship in the ad.
Work in English
English is required, per the ad.
Work on-site in Barcelona, on-site
No relocation package mentioned.
Have 3+ years of experience
Mid-level role.

Requirements

What we're looking for

What we need

Three or more years of professional experience, with at least two of them running projects or engagements end to end, where failure had a real cost and it did not fail. You can walk us through it with numbers.
A background where structure, pace and client exposure were the job: strategy consulting, banking or finance, a high-growth startup, or running client projects in a services business. What matters is that you carried the outcome rather than the analysis.
AI native, demonstrated rather than claimed. You use models daily in your own work and you can walk us through something you built with them, such as an automated reporting pass, a triage or sampling step, or a prompt chain that replaced manual work. You build what you need without being shown.
Experience coordinating large distributed groups of people you do not line-manage, or having built the system that let a small team run hundreds.
Comfort with numbers. You can take a margin, a throughput rate and a staffing plan and say what has to change.
Professional English, used daily with clients.

The role

A project at Sovrano can run with dozens of experts or with more than a thousand, and it has to scale without the work slipping. Someone owns that end to end, account by account: who is on the project, whether it ships on time and to standard, what the client hears, and what it costs. That is this seat. Sovrano AI sells expert capacity. Frontier AI models are no longer limited by compute. They are limited by how much verifiable expert human judgment can be pointed at them, and Europe has the experts without having the companies that put them to work. We recruit and verify domain experts through 100 or more partner universities across Europe, a network of 400,000 or more professionals covering 23 languages and holding EFMD certification, then put them to work on the data that trains and evaluates frontier models. Clients are frontier AI labs, robotics companies and enterprise AI teams across Europe and the United States. We also run our own European executive benchmark, EuroExec. The week is spent running live client work rather than planning it. Roughly two thirds of it is execution: the schedule, the squad, the throughput, the escalations. A quarter of it is the client: the weekly report, the call, and the conversation when something slips. The rest is the numbers. You will restructure a project when the throughput is not there, onboard hundreds of experts in days because that is the pace clients buy, sit in the client call when the news is bad, and take someone off a project when the work is not good enough. It is not a comfortable job. Volumes arrive quickly, deadlines are measured in days, and the client is on the phone either way. The people who do well here are energised by that rather than braced against it. Project Lead is the seat that runs the work. Special Project Lead is the step up from it: the same account ownership with the technical substance added, which means designing the quality systems themselves, authoring and rewriting rubrics, and holding a line with research on what a quality signal is measuring. Most people will grow into it from this role. Candidates who already have that technical depth can be considered for it directly.

What you'll do

What you own

The accounts you own, end to end: the plan, the squad, the schedule, the client relationship, the margin.
Scale. Taking a project from dozens of experts to hundreds or thousands is a structure problem, and you build the structure.
Squad performance once the experts are staffed, including who stays on a project and who comes off it.
The expert community on your projects. They stay briefed, engaged and producing. You set how the project communicates with them and you keep that running at volume.
Onboarding at pace. A new workstream can need hundreds of qualified people inside a week, and you own how that happens without the first work suffering for it.
The quality bar on live work. You apply the rubrics and the sampling method, you act on what they surface, and you say when they stop catching errors.
Client reporting on live work, and the truth about what is hard.
Escalations. They end with you, not above you.
Four numbers, posted every week: people staffed against what the client needs, hours completed, rework rate, margin per hour.