Senior Software Engineer – AI/HPC Middleware

Cornelis Networks develops high-performance scale-out networking solutions for AI and HPC datacenters, integrating hardware, software, and system-level technologies. The Senior Software Engineer will design, develop, optimize, and upstream-enable AI/HPC communication middleware over the company’s interconnects, while collaborating across the stack to improve performance, correctness, and deployability. The role also involves analyzing customer issues, validating performance, and contributing to open-source projects.

Responsibilities:

• AI/HPC Middleware Enablement & Optimization: Implement and help optimize features for HPC middleware (HPI and SHMEM) and AI middleware (CCL stacks ie: NCCL/RCCL and related collective communication libraries)
• Transport & Integration: Assist in integrating middleware capabilities with underlying transports and provider layers (ie: libfabric / OFI, UCX, verbs-like semantics where applicable)
• Upstream & Open-Source Leadership: Prepare and submit upstream contributions across MPI/SHMEM projects, CCL ecosystems, and related components
• Cross-Stack Collaboration & Performance Validation: Collaborate with kernel/driver and switch teams to help deliver end-to-end performance aligned to the Cornelis product roadmap
• Customer & Field Impact: Help analyze performance traces and reproduce customer issues, and assist in translating findings into fixes

Requirements:

• • 6+ years of experience in systems programming in C/C++ on Linux; Bachelor's degree in Computer Science, Engineering, or related field (or equivalent experience)
• • Hands-on experience with HPC middleware internals (Open MPI, MPICH, MVAPICH, and/or SHMEM) or optimizing collective communications for AI, including NCCL/RCCL, CUDA/ROCm) and/or other CCL stacks
• • Demonstrated ability to deliver low-latency/high-throughput communication paths and to diagnose performance issues using profiling/tracing tools
• • Working knowledge of transport/integration layers such as OFI/libfabric, UCX, verbs-style concepts
• • Solid background in RDMA concepts and performance tuning
• • Experience contributing to open-source projects
• Location: This is a remote position for employees residing within the United States
• • Prior experience developing or maintaining libfabric providers
• • Familiarity with Ultra Ethernet (UEC/UET) specifications
• • Experience with RoCEv2, congestion control, and/or Ethernet-based RDMA deployments
• • Experience with cluster-scale benchmarking, profiling, and optimization
• • Background with HPC fabrics, such as Omni-Path