Role Overview
Marvell is seeking a research intern to explore how future AI networks behave at scale. The role uses simulation to study scale-up and scale-out protocols, congestion, and interactions among workloads, fabrics, and hardware before silicon. The internship offers access to advanced infrastructure and mentorship from senior architects and researchers, with work intended for publication and product impact.
Responsibilities
- Investigate emerging scale-up and scale-out protocols under realistic AI workloads.
- Extend simulation platforms to model networking and congestion-control behavior at scale.
- Experiment with AI-scale and LLM-serving simulation, including workload/trace-driven modeling using tools such as Astra-Sim, Chakra, and htsim.
- Evaluate protocol performance, scalability, and resilience across topologies.
- Correlate simulation results with architectural and hardware behavior to generate insights.
Requirements
- Enrolled in a PhD, MS by Research, or MTech by Research program in Computer Science, Computer Engineering, Electronics, Electrical Engineering, or a related field.
- Strong fundamentals in computer networks, transport protocols, and distributed systems.
- Curiosity about large-scale simulation and frameworks such as Astra-Sim, Chakra, htsim, ns-3, gem5, or SST.
- Understanding of scale-up/scale-out fabrics, RDMA, or congestion control, or eagerness to learn.
- Programming experience in C++ and Python.