Nscale is a GPU cloud engineered for AI, providing infrastructure for AI start-ups and large enterprises. The Principal Network Engineer will lead the technical direction and operational strategy for Nscale's AI interconnect networks, focusing on reliability and scalability while mentoring other engineers.
Responsibilities:
- Owning the technical direction and operational strategy for Nscale's AI interconnect networks
- Designing, reviewing, and evolving large-scale Infiniband and RoCE fabric architectures to support future growth and workload demands
- Acting as the senior escalation point for the most complex network incidents, guiding deep technical investigations and systemic fixes
- Driving cross-team initiatives to improve fabric reliability, performance predictability, and operational maturity
- Defining standards for hardware configuration, congestion control, routing, firmware lifecycle management, and change safety
- Partnering with SRE, Compute Platform, and Network Architecture teams to influence end-to-end system design
- Mentoring senior and mid-level network engineers, raising the bar for operational rigor and technical excellence
- Driving measurable improvements in uptime, latency consistency, capacity efficiency, and incident reduction
Requirements:
- 10+ years of experience in network engineering, with deep focus on HPC, AI, or hyperscale data centre networking
- Expert-level operational and architectural experience with Infiniband and/or large-scale RoCE fabrics
- Deep understanding of RDMA internals, congestion management, and fabric-level failure modes
- Strong expertise in modern data centre routing and control planes (BGP, OSPF, ECMP)
- Proven ability to debug and resolve cross-layer issues spanning hardware, firmware, kernel, and application communication libraries
- Demonstrated ability to lead complex technical initiatives across teams without direct authority
- A systems-level mindset, balancing performance, reliability, scalability, and operational cost
- Extensive experience with NVIDIA/Mellanox networking platforms in production AI or HPC environments
- Deep familiarity with distributed training frameworks and GPU communication patterns
- Experience designing network observability systems for high-cardinality, high-throughput environments
- Prior experience influencing platform or infrastructure strategy at scale