N
Senior Software Architect, GPU Networking
Sin especificarTiempo completoSoftware ArchitectureSemiconductorsComputer Hardware
Design and prototype GPU networking software for large-scale AI infrastructure, combining hands-on development with system-level architecture work and cross-team collaboration.
Responsabilidades
- Contribute to the architecture and prototyping of GPU networking software for large-scale AI infrastructure.
- Analyze requirements, compare design alternatives, and document technical decisions and trade-offs.
- Build proofs of concept to evaluate interfaces, performance, resiliency, and software-hardware interactions.
- Collaborate with GPU, switch, firmware, driver, networking software, product, and architecture teams to advance concepts toward implementation.
- Develop expertise in the GPU networking software stack and take increasing ownership of defined architectural areas.
Requisitos
- Bachelor’s or master’s degree in Computer Science, Electrical Engineering, Computer Engineering, or a related field, or equivalent experience.
- At least 5 years of relevant software development experience in Linux environments.
- Strong software engineering fundamentals, including debugging, testing, data structures, concurrency, and operating-system concepts.
- Foundational knowledge of computer networking.
- Ability to reason about complex systems, interfaces, and design trade-offs, and to communicate technical ideas clearly.
- Ability to collaborate across teams, locations, and time zones.
Se valora
- Knowledge of Ethernet and IP datacenter protocols, management or telemetry protocols, GPU interconnects, Linux kernel or driver development, or network operating systems.
- Experience creating prototypes or system-level architecture documents.
- Familiarity with AI communication libraries or programming models such as NCCL, UCX, MPI, or CUDA.
- Knowledge of Kubernetes, containers, or cloud-native infrastructure.
- Experience analyzing performance, reliability, resiliency, or observability in complex software systems.