Paris · Clinical AI · Radiology

Arthur Thévenet

I build clinical AI for radiology, and the engineering organisations that ship it.

W 255 · L 128drag to window

Now

Software Engineering Leader, Clinical AI — DeepHealth

I have been CTO of Gleamer, a Paris-based radiology AI company, since 2025, and led it through its acquisition by RadNet in March 2026. Since then I have also led clinical AI engineering inside DeepHealth (RadNet’s digital branch): 50+ engineers across six teams, spread over Europe and the United States.

The model is rarely the hard part. The hard part is running it in hospitals, on networks we do not control, and shipping updates regularly without breaking anything.

  • Voice — ASR SaaS for radiology
  • Infrastructure
  • AI Device
  • Reporting & viewing integration
  • Orchestration
  • Data & insights

Built

Voice

I recently led Voice, our speech recognition SaaS for radiology: radiologists dictate, the report writes itself, and it has to be fast enough that nobody goes back to the old way. I took it from a POC to production and past $1M ARR in six months, on an architecture that keeps the user’s audio inside HDS-certified health data hosting.

It is the clearest example of what I actually do, which is take a research capability and turn it into a product a customer will pay for and a regulator will accept.

Scale

What that looks like

The scale the systems I’m responsible for run at.

  • 60Mexams a year
  • 3,000sites live
  • 50+countries
  • 20+FDA-cleared / CE-marked devices
  • 6engineering teams
  • $270Macquisition by RadNet, 2026

Roughly 1.9 exams a second. 0 since you opened this page.

Practice

Where I am useful

  1. Shipping fast inside a quality system

    Keeping a real release cadence while a quality system is watching. Traceability, verification and design controls built into how the team already works, instead of bolted on before an audit.

  2. Scaling an engineering organisation

    Growing a group from thirty to seventy engineers without losing the thing that made it good. Hiring at pace, splitting squads across time zones, turning senior engineers into first-time managers, and building internal mobility so people grow without leaving.

  3. Engineering through an acquisition

    Technical and legal integration of a European medtech into a US parent — platform consolidation, security and compliance due diligence, merging two engineering cultures without stalling the roadmap, and keeping the people who matter.

  4. Clinical AI infrastructure

    Multi-tenant GPU inference on Kubernetes and GCP, on-premise deployments into hospital networks, and HDS and GDPR-compliant health data architecture. A cloud bill in the millions of dollars a year, which makes the cost model a product decision rather than a finance one.

  5. Radiology product

    DICOM and PACS, viewers, AI orchestration layers, and dictation. What radiologists actually change their workflow for, and what they quietly switch off.

Availability

Arthur Thévenet

Advisory and speaking

I speak now and then about clinical AI, radiology and engineering leadership, and have a handful of advisory conversations a year. If something above overlaps with what you are working on, write to me.

Background

Before this

  • DeepHealth · RadNetSoftware Engineering Leader, Clinical AI2026–Paris
  • GleamerChief Technology Officer2025–Paris
  • 7SpeakingChief Technology Officer2023–25Paris
  • DashlaneEngineering Manager, then Senior Engineering Manager2020–23Paris
  • Smartbox GroupSenior Software Engineer, then Mobile & Integration Engineering Manager2016–20Dublin
  • CoffreoBack-end engineer2014–16Clermont-Ferrand