Network Architect

The company

Nebius AI is an AI cloud platform with one of the largest GPU capacities in Europe. Launched in November 2023, the Nebius AI platform provides high-end, training-optimized infrastructure for AI practitioners. As an NVIDIA preferred cloud service provider, Nebius AI offers a variety of NVIDIA GPUs for training and inference, as well as a set of tools for efficient multi-node training.

Nebius AI owns a data center in Finland, built from the ground up by the company’s R&D team and showcasing our commitment to sustainability. The data center is home to ISEG, the most powerful commercially available supercomputer in Europe and the 16th most powerful globally (Top 500 list, November 2023).

Nebius’s headquarters are in Amsterdam, Netherlands, with teams working out of R&D hubs across Europe and the Middle East.

Nebius AI is built with the talent of more than 500 highly skilled engineers with a proven track record in developing sophisticated cloud and ML solutions and designing cutting-edge hardware. This allows all the layers of the Nebius AI cloud – from hardware to UI – to be built in-house, distinctly differentiating Nebius AI from the majority of specialized clouds: Nebius customers get a true hyperscaler-cloud experience tailored for AI practitioners. 

 

The role

We are looking for an expert-level Network Engineer (Network Architect) to join our team to ensure the development of our own data center infrastructure and the backbone network, research and develop the future of network interconnect for high-scale datacenter fabric with GPU interconnect, high-speed storage, in-depth investigations of emerging issues on the network together with neighboring server infrastructure, and cloud overlay teams.

You’re welcome to work in our office in Amsterdam.

 

In this position, your responsibility will be to:

• Provide technical design, operations, and assistance to related cross-functional team projects, including internal R&D and hardware as a service team.
• Design, evaluate, and provide solutions for all-around network services in cloud overlay and underlay environments.
• Research and evaluate R&D network equipment in the internal lab with the server team and network development team; prepare MVP of network segments for deploying new services.
• Design and develop data center networks with thousands of server ports, InfiniBand-based or other kinds of interconnect for GPU clusters, and automation and management tools.
• Provide documentation for end-to-end traffic flow, including intermediate devices like host-based FW/overlay data plane/cloud gateway, transit firewall, multi-vendor datacenter fabric and datacenter edge, and backbone part.
• Develop tools for in-depth investigation, monitoring, and healing of scalable datacenter networks, including data and control planes across vm/cloud overlay and underlay.
• Contribute to the launch of new regions within our cloud and GPU platforms.

 

We expect you to have:

• Expert and above-average understanding and experience in MPLS with different vendors, routing and switching for DataCenter, routing stack including proprietary vendors and opensource suites FRR/BIRD/goBGP, SR MPLS+SRv6, traffic balancing/ECMP, internet traffic engineering, and cloud overlay network technologies.
• Experience in researching, preparing, and documenting the developed network design and solution.
• Experience in developing plans and testing network infrastructure and vendors for specific purposes, workloads, and runtime.
• Experience in diagnosing problems with the TCP/IPv4/v6/multilayer encapsulation protocol stack in a CLOS topology data center network end-to-end, including intermediate equipment (FW, LB/uplinks and backbone).
• Knowledge of the design and internal architecture of modern network equipment, including ASIC, Whitebox, QoS mechanisms, and the operation of combination functionality.
• Experience with multivendor environments with backbox and white box switches, open-source NOS.
• Working proficiency in English.

 

It would be an added bonus if you had:

• Working knowledge of the GPU Interconnect cluster with more than 2k GPU cards.
• Practical experience with using SDK to Broadcom/spectrum ASICs. 

 

Does all that sound like your kind of challenge? Then join us!