kandidatu France

MLOps Engineer

Siège de l'AP-HP · Public hospitals

En français : Ingénieur MLOps

This job in plain words

You size and deploy a GPU cluster on OVHcloud, then run inference servers for language models. The job is based in Paris and focuses on keeping health data secure.

Open to non-civil servants

Can you apply?

The page includes an indicative pay range for contract staff, so contract staff can be hired. No other conditions are stated.

Diploma
Master's degree (bac+5) · Engineering degree or equivalent (computer science, mathematics, public health, etc.)
Experience
At least 3 years in infrastructure, DevOps or MLOps, including GPU workloads.

The profile they want:

  • Engineering degree or equivalent (computer science, mathematics, public health).
  • At least 3 years in infrastructure, DevOps or MLOps, including GPU workloads.
  • Real hands-on practice of Terraform and Ansible.
  • GPU workload deployment and Kubernetes in real conditions.
  • Good knowledge of open weights models and their memory sizing.
  • Solid Python, shell, APIs and containers.

What you would do

  • Size and deploy the GPU cluster on OVHcloud, described in Terraform.
  • Industrialise the GPU layer on Kubernetes: drivers, CUDA, card sharing.
  • Run inference servers and optimise model latency and throughput.
  • Expose an internal LLM gateway with authentication, quotas and usage tracking.
  • Guarantee isolation of health data and traceability of access.
  • Monitor and control GPU resource costs.

How to apply

Apply on the employer's website, reached from the official offer on Choisir le Service Public.

Good to know:

  • Job based in Paris (75).
  • Remote work: not stated in the offer.

We wrote this summary automatically from the official offer. Only the offer published on Choisir le Service Public counts: read it before you apply. Source: Choisir le Service Public (DGAFP), Etalab 2.0 licence.

Similar jobs