NVIDIA Switzerland AG
Principal Software Architect, GPU Networking Research(m/w/x)
Vollzeitnur vor OrtSenior
Zürich
Neuer Job?Nejo!
Dein persönlicher KI-Karriere-Agent
NVNVIDIA
Senior GPU Networking Architect(m/w/x)
Zürich
VollzeitVor OrtSenior
AI/ML
Nejo KI-Zusammenfassung
Optimizing GPU communication kernels for AI acceleration. 5+ years CUDA programming and GPU architecture knowledge required. Hybrid work, 4-day work week.
Anforderungen
- 5+ years CUDA programming, including writing and optimizing GPU kernels
- M.Sc. or equivalent in computer science, computer engineering, or related field
- Strong understanding of GPU architecture fundamentals
- Experience with systems-level C/C++ development in performance-critical environments
- Familiarity with GPU data movement mechanisms
- Ability to read and reason about GPU performance profiles
- Strong collaboration skills in multi-national, interdisciplinary environment
- Experience developing or optimizing communication kernels in NCCL, NVSHMEM, or similar
- Understanding of distributed deep learning parallelism techniques
- Background in RDMA, InfiniBand, high-speed networking, and GPU system topology
- Experience with overlap techniques to hide communication latency
- Proven experience evaluating and optimizing large-scale LLM training or inference workloads
Aufgaben
- Build and optimize GPU communication kernels
- Implement GPU communication kernels
- Optimize GPU communication kernels
- Improve kernel efficiency using GPU architecture knowledge
- Minimize latency in GPU kernels
- Overlap computation with communication
- Develop GPU-resident communication primitives
- Develop device-side APIs
- Profile GPU kernels end-to-end
- Tune GPU kernels end-to-end
- Identify compute, memory, and network bottlenecks
- Drive targeted GPU kernel optimizations
- Co-design communication strategies with teams
- Build proofs-of-concept for communication strategies
- Conduct experiments for communication strategies
- Perform quantitative modeling for communication strategies
- Contribute to evolving programming models
- Expose GPU-aware networking capabilities
Berufserfahrung
- 5 Jahre
Ausbildung
- Master-Abschluss
Sprachen
- Englisch – verhandlungssicher
Tools & Technologien
- CUDA
- C/C++
- GPUDirect RDMA
- Nsight Compute
- Nsight Systems
- NCCL
- NVSHMEM
- RDMA
- InfiniBand
- NVLink
- NVSwitch
- PCIe
- PyTorch
- TensorRT-LLM
- vLLM
Die Originalanzeige dieses Stellenangebotes in der aktuellsten Version findest du hier. Nejo hat diesen Job automatisch von der Website des Unternehmens NVIDIA erfasst und die Informationen auf Nejo mit Hilfe von KI für dich aufbereitet. Trotz sorgfältiger Analyse können einzelne Informationen unvollständig oder ungenau sein. Bitte prüfe immer alle Angaben in der Originalanzeige! Inhalte und Urheberrechte der Originalanzeige liegen beim ausschreibenden Unternehmen.
Noch nicht perfekt?
- NVIDIA Switzerland AG
Senior HPC and AI Network Software Architect(m/w/x)
Vollzeitnur vor OrtSeniorZürich - NVIDIA
HPC and AI Software Architect(m/w/x)
Vollzeitnur vor OrtBerufserfahrenZürich - NVIDIA Switzerland AG
HPC and AI Software Architecture Intern(m/w/x)
VollzeitPraktikumnur vor OrtZürich - NVIDIA
Senior Software Developer(m/w/x)
Vollzeitnur vor OrtSeniorZürich
NVNVIDIA
Senior GPU Networking Architect(m/w/x)
Zürich
VollzeitVor OrtSenior
AI/ML
Nejo KI-Zusammenfassung
Optimizing GPU communication kernels for AI acceleration. 5+ years CUDA programming and GPU architecture knowledge required. Hybrid work, 4-day work week.
Anforderungen
- 5+ years CUDA programming, including writing and optimizing GPU kernels
- M.Sc. or equivalent in computer science, computer engineering, or related field
- Strong understanding of GPU architecture fundamentals
- Experience with systems-level C/C++ development in performance-critical environments
- Familiarity with GPU data movement mechanisms
- Ability to read and reason about GPU performance profiles
- Strong collaboration skills in multi-national, interdisciplinary environment
- Experience developing or optimizing communication kernels in NCCL, NVSHMEM, or similar
- Understanding of distributed deep learning parallelism techniques
- Background in RDMA, InfiniBand, high-speed networking, and GPU system topology
- Experience with overlap techniques to hide communication latency
- Proven experience evaluating and optimizing large-scale LLM training or inference workloads
Aufgaben
- Build and optimize GPU communication kernels
- Implement GPU communication kernels
- Optimize GPU communication kernels
- Improve kernel efficiency using GPU architecture knowledge
- Minimize latency in GPU kernels
- Overlap computation with communication
- Develop GPU-resident communication primitives
- Develop device-side APIs
- Profile GPU kernels end-to-end
- Tune GPU kernels end-to-end
- Identify compute, memory, and network bottlenecks
- Drive targeted GPU kernel optimizations
- Co-design communication strategies with teams
- Build proofs-of-concept for communication strategies
- Conduct experiments for communication strategies
- Perform quantitative modeling for communication strategies
- Contribute to evolving programming models
- Expose GPU-aware networking capabilities
Berufserfahrung
- 5 Jahre
Ausbildung
- Master-Abschluss
Sprachen
- Englisch – verhandlungssicher
Tools & Technologien
- CUDA
- C/C++
- GPUDirect RDMA
- Nsight Compute
- Nsight Systems
- NCCL
- NVSHMEM
- RDMA
- InfiniBand
- NVLink
- NVSwitch
- PCIe
- PyTorch
- TensorRT-LLM
- vLLM
Die Originalanzeige dieses Stellenangebotes in der aktuellsten Version findest du hier. Nejo hat diesen Job automatisch von der Website des Unternehmens NVIDIA erfasst und die Informationen auf Nejo mit Hilfe von KI für dich aufbereitet. Trotz sorgfältiger Analyse können einzelne Informationen unvollständig oder ungenau sein. Bitte prüfe immer alle Angaben in der Originalanzeige! Inhalte und Urheberrechte der Originalanzeige liegen beim ausschreibenden Unternehmen.
Über das Unternehmen
NVIDIA
Branche
IT
Beschreibung
The company is developing groundbreaking solutions in Virtual Reality, Artificial Intelligence, Deep Learning, and Autonomous Vehicles.
Noch nicht perfekt?
- NVIDIA Switzerland AG
Principal Software Architect, GPU Networking Research(m/w/x)
Vollzeitnur vor OrtSeniorZürich - NVIDIA Switzerland AG
Senior HPC and AI Network Software Architect(m/w/x)
Vollzeitnur vor OrtSeniorZürich - NVIDIA
HPC and AI Software Architect(m/w/x)
Vollzeitnur vor OrtBerufserfahrenZürich - NVIDIA Switzerland AG
HPC and AI Software Architecture Intern(m/w/x)
VollzeitPraktikumnur vor OrtZürich - NVIDIA
Senior Software Developer(m/w/x)
Vollzeitnur vor OrtSeniorZürich