Infrastructure Edge-AI and Cloud

Project overview

We are looking for an Infrastructure Edge-AI and Cloud specialist for a 20-day mission with possible renewal.

PART-TIME MISSION

CONTEXT

MISSIONS

Lot 0 – Management and governance, cross-functional (approx. 3 days)

Priority 1 – Paris bay operational (approx. 5 days)

Priority 1 – Supervision and security (approx. 4 days)

Priority 2 – Marseille bay (approx. 2 days)

Documentation and transfer (approx. 2 days)

REQUIRED SKILLS

PROFILE

• Intelligence Lab: Pole program: Edge-AI infrastructure on campuses (DGX Spark, Jetson, Raspberry Pi bays) coupled with an OVH cloud platform (coding assistant on open-weight models), managed from the portal.
• Deployment engaged: Paris bay in service (upgrade to be done), network hardware and NVIDIA workstations already delivered to Marseille, Bordeaux, and Lyon. OVHcloud selected on 09/04 (2 x H100 + 2 x L40S, embeddings, LiteLLM VM, Gravelines region, Oct. 2026
• June 2027). Qwen3 27B model benchmarked, replaceable.
• OVHcloud "Cloud for Education" Partnership: To be signed, €100k excl. tax credits for ~4,000 beneficiaries, distributed in batches of codes via teachers and the Intelligence Lab Manager. Student pseudonymization required by specifications (FC-25 requirement), access via API keys.
• Departure of the current technical project manager: Take over and finalize the deployment with the existing team: two students handling run and on-site installation in Marseille and Lyon, the IT department, campus referents. No MCO provider.
• Priorities: 1) Paris bay and cloud platform operational, 2) Marseille, Bordeaux, and Lyon bays in optional phase depending on progress.
• Objective: Deliver an operational end-to-end infrastructure (edge, cloud, portal), equipped with its supervision, consumption monitoring, and governance tools, documented and transferable (group platform managed, possible takeover by IT department in 1-2 years).
• Out of scope: Room setup, furniture, signage, painting, and physical logistics of the campuses. No bay in Rennes.
• Take over the deliverable backlog, set the target schedule by priority and the RACI of contributors.
• Lead the weekly meeting, maintain the risk and decision log, produce bi-monthly reporting to the Requester and Management.
• Work in tandem with the Intelligence Lab Manager (designated OVHcloud contact, recruitment in progress) and with the IT department. Technically supervise the two students handling the run.
• Propose the target MCO model (provider, IT department, or school team) with its run RACI and cost, for decision before the end of the mission.
• Upgrade the existing infrastructure (DGX Spark, Jetson Orin Nano Super, Jetson Thor, Raspberry Pi). Deploy Docker and Kubernetes on the bay machines.
• Build the unified supervision dashboard, designed for multi-campus (site-to-site VPN) and put into service in Paris.
• Deploy the Fortinet VM and log export to IT department servers.
• Monitor Eaton UPS (SNMP cards, probes). Set up alerts for temperature/fire/smoke/water leaks with automatic machine shutdown and notification to general services.
• Obtain prerequisites from the IT department and campus referent: dedicated internet connection to the bay (≥ 1 G or unfiltered IP, no date yet), electricity, air conditioning, RJ45 cabling.
• Commission the DGX Spark bay (UPS, switch, probes) based on the Paris reference configuration. Connect to centralized supervision.
• Test the local computing environment with the campus referent.
• Write technical documentation (edge, cloud, network, supervision) and operating procedures at the level required for takeover by an MCO provider or the IT department.
• Produce student and teacher tutorials. Ensure skills transfer to school teams and IT department.
• Containerization and orchestration: Docker, Kubernetes (bare-metal / K3s), PXE boot, Linux, ARM architectures (Jetson, Raspberry Pi).
• GPU and AI Infrastructure: NVIDIA DGX / Jetson, CUDA, open-weight LLM serving (vLLM, LiteLLM), API key and quota management.
• Cloud: OVH Cloud (GPU instances, storage), remote backups, Infrastructure as Code.
• Network and security: Site-to-site VPN, VLAN, Fortinet, log centralization, access management.
• Supervision: Grafana / Prometheus or equivalent, UPS monitoring (SNMP / NUT), environmental alerting.
• Integration: REST API, integration with an existing web portal, Python / Bash scripting.
• Management: Multi-site coordination, supervision of junior contributors, reporting, documentation.
• Senior DevOps / Infrastructure Engineer (5+ years) with experience as a tech lead or technical project manager on multi-site deployments.
• End-to-end GPU / AI deployments on-premise and cloud, ideally in an educational or research environment.
• Autonomous and pragmatic: take over an ongoing project, make decisions, close out.

Project details

• Start: 2026-10-19
• Duration: 20 jours ouvrés (renouvelables)
• Location: 92 - LA DEFENSE, FR
• Languages: French