Our team is supporting the development and implementation of Google Cloud Graphics Processing Unit (GPU) roadmap, including the integration of shared capabilities to enhance Machine Learning (ML) fungibility. We have a robust pipeline of programs scheduled for General Availability in 2026, including: A4X+ and A5.
We're not just maintaining the status quo. We're innovators delivering novel capabilities such as Graphics Processing Unit Remote Direct Memory Access (GPU RDMA), ML Networking for Virtual Machines (VMs) and bare metal, monitoring, packet telemetry, etc. Our team is growing rapidly, and with you on board, it will grow even stronger. As we deliver these key projects, you will have a unique opportunity to shape the future of Google's network. Your ideas and expertise will directly influence how we build and evolve the infrastructure that powers everything at Google.
Responsibilities
- Understand capabilities provided by series of ConnectX (CX) network interface cards (NICs).
- Design features that integrate graphics processing unit (GPU)-to-GPU communication capabilities into the Google Cloud infrastructure.
- Code and implement the features that enable GPU-to-GPU communication on virtual machine (VM) families.
- Deliver virtual machine learning (ML) networking infrastructure enabling ML workloads to run in Google Cloud Platform (GCP).
- Enable GPU remote direct memory access (RDMA) networking for VMs and bare metal by exposing the NICs.
Qualifications
Minimum qualifications:
- Bachelor's degree or equivalent practical experience.
- 5 years of experience with one or more general purpose programming languages including but not limited to: Java, C/C++, Python, or Go.
Preferred qualifications:
- Experience with software architecture, software engineering, networking protocols, network virtualization, or networking.