Note: The job is a remote job and is open to candidates in USA. RoShay Services is seeking a highly experienced Senior Software Engineer to design, develop, and upstream-enable AI communication middleware for distributed AI workloads. The role involves implementing performance-critical features, optimizing distributed training, and collaborating with cross-disciplinary teams for system optimization.
Responsibilities
- Implementing performance-critical features for fabric enablement and optimizing distributed training across multi-node, multi-GPU systems
- Enhancing GPU communication paths, including GPU-direct transfers, IPC, and synchronization
- Profiling workloads to identify bottlenecks and tuning AI frameworks such as PyTorch Distributed, TensorFlow/XLA, and JAX
- Developing benchmarks, contributing to open-source projects, and collaborating with cross-disciplinary teams for system optimization
- Prototyping Ultra Ethernet capabilities and providing deployment and performance validation insights
- Supporting troubleshooting and performance analysis through trace analysis and robust fixes
Skills
- 8+ years of experience in high-performance systems programming in C/C++ on Linux
- Strong expertise in GPU communication stacks (CUDA/ROCm, NCCL/RCCL)
- Experience optimizing distributed training performance, understanding collective communication, and delivering production-quality code
- Open-source contributions in relevant areas
- Preferred experience with AI frameworks, libfabric/OFI, UCX, RDMA, RoCEv2, and cluster performance testing
Benefits
- Comprehensive rewards package including salary, incentives, health benefits, and flexible work options
Company Overview