Tabular AI Solution Architect

Probabl · United States · posted today ago

engineering machine-learning remote jobs in united states

Listing supplied by Himalayas. 101 Careers did not originate this post and applications are handled by the employer.

About this role

The role

You are the technical voice of Probabl in front of US data science teams: the person who takes a first conversation to a technical discovery, a demo on their own problem and a scoped proof of concept, alongside the sales team. Services owns technical pre-sales and delivery; you bring that field engineering expertise to the sales team on every US opportunity.

Probabl is the Tabular AI company founded by the creators of scikit-learn. Skore, our platform, builds, validates and tracks predictive models on tabular data with the methodology of the scikit-learn maintainers built in. Data science is being rewritten by agents faster than the discipline can absorb it; our job is to keep the science guarding the speed.

This is our first Solution Architect in the US. You join the Services team, Probabl's applied AI team of field engineers and solution architects, and work day to day with our US sales lead. Our pre-sales practice already runs in France, with a Solution Architect, a reference demo, a demo kit and a PoC playbook. You run it for the American market and feed back what you learn, so the whole team's practice gets better.

What you'll do

Run the demo.Own the reference Skore demo for the US and adapt it to a prospect's use case in days, with the demo kit, on your own.
Run technical discovery. With data science leads and their platform teams: current stack and tooling, how models are built and validated today, where reliability leaks, what a first project would look like.
Scope the PoC.Turn a qualified opportunity into one scoped question on the customer's data, with a clear go/no-go, then hand it to our field engineers to deliver.
Run workshops and hackathons. Hands-on sessions with a prospect's data science team on their own data, and the occasional hackathon with partners or schools: the fastest way to show what Skore does and to qualify a team.
Answer the hard questions. Architecture, security, deployment options (on-prem, BYOC, cloud), how Skore fits an existing stack. You bring Engineering in when needed, not by default.
Work the deal with Sales.Technical qualification, RFPs and evaluation workshops, alongside our US sales lead.
Bring the field back to Product. What US customers ask for, what blocks a deal, what wins it: written up, sized, and fed into the roadmap every week.

What you bring

5+ years in data science or ML engineering, then pre-sales or solutions work for a data or AI product. You have built models in production and you have sold the idea of building them better.
You master tabular data work end to end, whatever the framework: scikit-learn of course, but just as much gradient boosting, PyTorch, time series and the like, with the evaluation and validation that go with them. You can read a prospect's pipeline, spot the leakage, and show them in a notebook what a sound evaluation looks like.
You run a room. A live demo to 20 data scientists, a whiteboard with a platform team, a discovery call with a VP: you adjust and you land the point.
You are comfortable with enterprise buying groups: data science, IT and security, procurement, and the economic buyer, each with their own question.
Based in the US, remote, with travel to customers and periodic trips to Paris.
Strong plus: experience with agentic or LLM-based data science tooling, and a healthy skepticism about its outputs.
Nice to have: Kubernetes and cloud basics, tabular and time series foundation models (TabICL and others), an open-source footprint in the Python data ecosystem.

Why Probabl

Credibility you don't have to build. scikit-learn is downloaded over 200 million times a month. You walk into every room with the stewards of the library behind you.
A category, not a feature. We are defining Tabular AI: AI on the structured business data that runs companies, with the scientific method built in. You help define what it means in the US.
A team that builds its playbook together.We have a first version of our demos, PoC playbook and processes, and they still have a lot of room to evolve. You use them with US customers and improve them with the rest of the team.
Short distance to the product. Solution Architects and field engineers sit in a weekly Product–Services sync. What you hear from customers changes the roadmap, visibly.

Practical

Reports to the VP Services (Paris). Day to day with our US sales lead, who is fully involved in the hiring.
Location: United States, remote. Preferred time zone: East Coast (to confirm). Travel to customers as needed; two to three trips a year to Paris, including onboarding.
Compensation:fixed + variable tied to pipeline and closed deals; band to confirm.
Start:as soon as possible.
Process, three weeks from first call to offer: