Accelerated Computing Solutions Analyst - Early Career

Job Title

Accelerated Computing Solutions Analyst - Early Career

Role Summary

The Accelerated Computing Solutions Analyst will research and build co-optimized hardware/software solutions for AI data-center infrastructure, focusing on disaggregated memory, storage, networking, and accelerator architectures. The role sits on the Technology and Solutions Architecture team and collaborates with architects, product and engineering teams to define and evaluate next-generation, high-performance AI/ML system designs.

This is an early-career technical role contributing to system architecture, performance characterization, proof-of-concept development, and standards engagement.

Experience Level

Entry-level / Early Career. (Specific years of experience not specified.)

Responsibilities

Primary responsibilities include system architecture, evaluation, and prototype execution for data-centric AI infrastructure.

• Partner with hardware and software architects and IC engineering teams to design scalable AI/ML infrastructure and co-optimized solutions.
• Lead system architecture design and implement proof-of-concept solutions for heterogeneous and disaggregated compute, memory, and storage environments.
• Perform advanced workload characterization and performance analysis (latency, response time, resource utilization) for AI data processing.
• Develop and refine simulation and modeling environments for large-scale AI/ML setups.
• Advocate and represent the company in industry standards bodies and working groups.
• Create documentation and data visualizations to communicate findings and support decision-making.
• Monitor developments in generative AI and hardware data analysis to propose improvements to device tuning and interoperability.

Requirements

Must-have technical skills and experience; nice-to-have items listed separately.

• Must-have: HW/SW co-design experience with GPUs, DPUs, FPGAs, or custom hardware accelerators.
• Must-have: Knowledge of processor micro-architectures (e.g. ARM), scalar/vector pipelines, and techniques to optimize memory-bound applications.
• Must-have: Deep understanding of memory and storage subsystems, including disaggregated memory approaches and NVMe over Fabrics.
• Must-have: Familiarity with coherent and non-coherent interconnects such as CXL, UAL, NVLink, and multi-GPU / multi-node communication and optimization techniques (e.g. NCCL, SHARP).
• Must-have: Proficiency in performance benchmarking, system profiling, and end-to-end analysis of AI data workloads.
• Must-have: Experience with simulators and modeling for system-level evaluation and GPU/CPU cooperative accelerated computing.
• Nice-to-have: Experience with low-latency transport protocols, post-quantum cryptography, and working in standards bodies.
• Nice-to-have: Proficiency with Python and data visualization/tools for documenting analysis and results.

Education Requirements

PhD or Master's degree in Computer Science and Engineering, Electrical Engineering, or a related technical field.

About the Company

Company: Marvell Technology

Headquarters: Santa Clara, California, United States

Marvell’s semiconductor solutions serve as essential building blocks of the data infrastructure connecting our world, driving innovation across enterprise, cloud, AI, and carrier architectures. The company focuses on creating transformative technology that shapes the future.