Datacenter HPC Fabric Engineer for GPU & Storage

Stealth Startup AI infrastructure; Lombardia; Sur place; EUR 40 000 - 95 000; Plein temps

Avantages offerts par ce poste

Equity

Résumé du poste

Stealth Startup AI infrastructure in Milan, Italy, seeks a Datacenter & HPC Fabric Engineer to design, build, and tune an intra-pod fabric for a high-performance compute unit.

You will turn a 100GbE lab fabric into a production design, selecting switches, NICs, optics, and cabling, and delivering documented performance, configuration-as-code, and BOM-ready outputs for series production.

Qualifications

• Bachelor’s degree in Electrical Engineering, Computer Science, Networking, or equivalent field.
• 3-8 years of network engineering in datacenter, HPC, or cloud environments.
• Hands-on experience designing, building, and operating datacenter fabrics.
• Exposure to performance-sensitive networking (HPC, AI/GPU, or storage), or a demonstrated ability to ramp up quickly.
• Comfort with hands‑on lab and physical work, from server networking to racking and cabling.
• English speaking.

Responsabilités

• Design and build the intra-pod fabric, and select switches, NICs, optics, and cabling.
• Tune the fabric for low-latency, high-throughput GPU and storage traffic, and benchmark it.
• Make the network capacity available to GPU workloads and virtual machines with minimal overhead.
• Design the storage and out-of-band management networks.
• Work with electrical and mechanical engineers on port layout, cabling, and serviceability.
• Produce port maps, cabling and BOM inputs, and test reports.

Connaissances

Datacenter fabrics

Performance-sensitive networking

Hands-on lab work

English speaking

GPU/AI cluster experience

RDMA tuning

Open networking

Hardware startup experience

Formation

Bachelor’s degree in Electrical Engineering, Computer Science, Networking, or equivalent field

Description du poste

Stealth Startup AI infrastructure in Milan, Italy, seeks a Datacenter & HPC Fabric Engineer to design, build, and tune an intra-pod fabric for a high-performance compute unit.

You will turn a 100GbE lab fabric into a production design, selecting switches, NICs, optics, and cabling, and delivering documented performance, configuration-as-code, and BOM-ready outputs for series production.