About the role#
This PhD internship focuses on high-speed networking and RDMA systems to support AI and LLM infrastructure. You will work on designing and deploying transport protocols and congestion control algorithms at scale. The role involves co-designing host, network, and application systems to improve performance and reliability. You will also track industry and academic trends to identify research opportunities and contribute to academic papers.
What you'll need#
- Current enrollment in a PhD program for computer networking or a related technical field.
- Proficiency in C, C++, Python, or Go.
- Familiarity with network protocols like TCP and RoCEv2, plus experience with Socket and verbs APIs.
- Understanding of data center congestion control algorithms.
- Knowledge of scale-up protocols such as PCIe, NVLink, and UALink.
- Familiarity with high-speed network systems including RDMA and AI network optimization.
- Basic knowledge of GPU architecture.
- Experience with high-performance communication frameworks like NCCL, MPI, or RPC libraries is preferred.
- Familiarity with AI training and inference systems is a plus.
- Publications in top-tier networking conferences such as NSDI, SIGCOMM, OSDI, or SOSP are preferred.
What you'll do#
- Design, optimize, and deploy high-performance transport protocols and congestion control algorithms for AI and LLM workloads.
- Research and develop AI communication frameworks and network protocol stacks.
- Improve the scalability and performance of AI networks through host-network-application co-design.
- Identify innovative system components and present findings in academic publications.
Location & details#
- Location: San Jose, California.
- Modality: On-site.
- Employment: Full-time internship.
- Hiring process: Rolling basis.


