Principal engineer – GPU Memory Systems & Scale-Up/Out Systems

METAVERSE COMPUTING LIMITED; Hong Kong; On-site; HKD 1,200,000 - 1,600,000; Full time

Job summary

METAVERSE COMPUTING LIMITED is seeking a Principal engineer to push the boundaries of GPU memory systems and scale-up/out architectures. You will design memory hierarchies, interconnects, and distributed systems for AI/ML and HPC workloads, validating ideas with simulators or hardware prototypes.

The role requires a PhD or equivalent, deep expertise in computer architecture and memory systems, and a track record of publications in prestigious venues.

Qualifications

• PhD in CS/EE or related field.
• (or equivalent research experience).
• Strong background in computer architecture, memory systems, and parallel/distributed computing.
• Hands-on experience with GPU architectures, CUDA, RDMA, or high-performance interconnects (CXL).
• Proficiency in performance modeling and simulation.
• Strong programming skills in C/C++, Python, or SystemVerilog/HDL.
• Track record of publications in ISCA, MICRO, ASPLOS, HPCA, SC, or related venues.

Responsibilities

• Research and develop novel GPU memory architectures (cache hierarchies, near-memory computing, disaggregated memory, link/switch optimizations).
• Design and evaluate scale-up and scale-out systems for GPU clusters, focusing on interconnect topologies, communication protocols, and load balancing.
• Optimize memory bandwidth, latency, and capacity utilization for large-scale GPU workloads.
• Collaborate with hardware and software teams to prototype new architectures (simulators, FPGA emulation, or real hardware).
• Publish cutting-edge research in top conferences/journals and contribute to patents.
• Engage with industry and academic partners to drive innovation in GPU-accelerated computing.

Skills

GPU architectures

CUDA

RDMA

CXL

Performance modeling

Parallel computing

Distributed computing

SystemVerilog/HDL

Python

C/C++

Education

PhD in Computer Science, Electrical Engineering, or related field

Tools

GPGPU-Sim

Gem5

NS3

Job description

Principal engineer – GPU Memory Systems & Scale-Up/Out Systems

We are seeking a highly motivated **Principal engineer** to advance the state-of-the-art in **GPU memory systems** and **scale-up/out computing architectures**. This role involves designing and optimizing memory hierarchies, interconnects, and distributed systems to improve performance, efficiency, and scalability for next-generation GPU-accelerated workloads (e.g., AI/ML, HPC, and large-scale data analytics).

The ideal candidate will have deep expertise in **computer architecture, memory systems, parallel computing, and distributed systems**, with a strong publication record in top-tier conferences (e.g., ISCA, MICRO, ASPLOS, HPCA, SC, NeurIPS).

Key Responsibilities

• Research and develop novel **GPU memory architectures** (e.g., cache hierarchies, near-memory computing, disaggregated memory, link/switch optimizations).


• Design and evaluate **scale-up and scale-out systems** for GPU clusters, focusing on **interconnect topologies, communication protocols, and load balancing**.


• Optimize **memory bandwidth, latency, and capacity utilization** for large-scale GPU workloads.


• Collaborate with hardware and software teams to prototype new architectures (e.g., using simulators, FPGA emulation, or real hardware).


• Publish cutting-edge research in top conferences/journals and contribute to patents.


• Engage with industry and academic partners to drive innovation in GPU-accelerated computing.

Required Qualifications

• **PhD in Computer Science, Electrical Engineering, or related field** (or equivalent research experience).


• Strong background in **computer architecture, memory systems, and parallel/distributed computing**.


• Hands-on experience with **GPU architectures, CUDA, RDMA, or high-performance interconnects (CXL)**.


• Proficiency in **performance modeling and simulation** (e.g., GPGPU-Sim, SST, Gem5, NS3).


• Strong programming skills in **C/C++, Python, or SystemVerilog/HDL** for prototyping.


• Track record of publications in **ISCA, MICRO, ASPLOS, HPCA, SC, or related venues**.

Preferred Qualifications

• Experience with **disaggregated memory, cache coherence protocols, or memory-centric computing**.


• Knowledge of **AI/ML workloads and their memory/system bottlenecks**.


• Contributions to **open-source hardware/software projects** (e.g., LLVM, PyTorch, OpenMPI).


• Familiarity with **datacenter-scale systems** (e.g., distributed training, high-performance storage).

Why Join Us?

• Work on **cutting-edge research** with real-world impact in AI, HPC, and cloud computing.


• Collaborate with **world-class researchers and engineers**.


• Competitive compensation, equity, and publication/patent incentives.

This job description balances technical depth with broad applicability, attracting candidates with expertise in **GPU memory systems** and **large-scale computing**. Let me know if you'd like any refinements!