Expertise · AI infrastructure recruiting
Network and Fabric Recruiting for AI Training Environments
We recruit fabric engineers and architects for AI training and inference: InfiniBand, RoCE and high-performance Ethernet, RDMA, congestion control, topology design and network automation. Most enterprise and campus network engineers haven't worked with collective communication at this scale, so we source and screen for it separately.
Audience
We work with GPU cloud platforms, AI labs, HPC centers, hyperscale operators and infrastructure vendors building or supporting training fabrics.
Common hiring problems
- Training jobs slow down and no one can separate a fabric problem from a job design problem.
- Most applicants are enterprise network engineers with no RDMA or collective communication experience.
- Vendors are making fabric design decisions because the internal architect role is still open.
- Automation and validation are behind the build, so every fabric change is made by hand.
Role families
Roles and screening
Fabric architecture
Fabric architect, network architect for AI clusters
Topology choice, oversubscription decisions, and the tradeoffs behind InfiniBand versus high-performance Ethernet in a specific build.
Fabric engineering and operations
InfiniBand engineer, RoCE engineer, high-performance network engineer
Subnet management, congestion control, link and cable fault isolation, and validating a fabric before jobs land on it.
Network automation
Network automation engineer, network reliability engineer
Configuration as code, validation pipelines and limiting the impact of changes on live fabrics.
Data center networking
Data center network engineer, edge and interconnect engineer
Covers the campus, backbone and interconnect layers that carry storage and north-south traffic around the cluster.
Environments
Technologies and environments
- InfiniBand with subnet managers and adaptive routing
- RoCEv2, PFC and ECN based lossless Ethernet
- Rail-optimized and fat-tree topologies
- NCCL and collective communication behavior
- Automation with configuration as code and validation pipelines
Role scope
Requirement mistakes
Sort these out before outreach starts.
- Screening on switch vendor familiarity without asking how the candidate's fabrics behaved under collective load.
- Treating high-performance fabric work as a subset of enterprise network engineering.
- Assigning fabric validation to a team with no visibility into job performance.
- Hiring one generalist for architecture, operations and automation at the same time.
Process
Search process
Czarina Tabayoyong agrees on the scorecard with your interview panel, builds a list of named candidates and contacts them directly. You get a written report every week. At the end, your team keeps all the research and the candidate pipeline. We use AI tools for research, and Czarina makes the recommendations herself.
Talent Intelligence Sprint
A fixed-scope research project on the candidate market before you commit to a search. Scopes are listed on the homepage.
Embedded Search Sprint
A recruiter inside your process for a defined sprint.
Site Launch or Team Build
Monthly recruiting support timed to your project milestones.
FAQ
Recruiting questions
- Why is fabric hiring harder than general network hiring?
- There are fewer candidates, and most of them work for a small number of employers. We ask them to describe how they isolated a link fault that only showed up as slow training, or how congestion control settings changed job completion time.
- Do you recruit for InfiniBand and Ethernet fabrics?
- Both. Many organizations run InfiniBand for training and high-performance Ethernet elsewhere. Experience with lossless networking, congestion control and topology design carries over between the two.
- Which adjacent markets do you map?
- HPC centers, research computing, financial low-latency networking, telecommunications and hyperscale network organizations. Candidates from each can move into AI fabric work with some ramp-up time.
- Can you support a vendor-side team?
- Yes. Infrastructure vendors need field, solutions and support engineers who can talk through a fabric with a customer. We screen them differently from internal operations hires.
GPU and HPC recruiting
Cluster engineering, fleet operations and scheduling around the fabric.
AI infrastructure recruiting
How the layers fit together and what to hire first.
Request a talent market map
How many people do this work in each metro, and which employers they work for, before you open the role.
Next step
Bring the scorecard and the deadline. We'll tell you what the candidate market looks like and which engagement we'd recommend, including when your team can fill the role internally.