Bachelor's or master's degree in computer science, Electrical Engineering, Computational Science, or related field; advanced degrees or certifications are a plus.
3+ years of experience in HPC development with a focus on parallel and distributed computing.
5+ years of experience in scientific software development.
Proficiency in CUDA and familiarity with NVIDIA's HPC software stack (cuDNN, NCCL, TensorRT).
Expertise in MPI, OpenMP, and other parallel programming paradigms.
Strong proficiency in C/C++, Python, and tools like Ansible/Terraform/Packer.
Familiarity with profiling and performance optimization tools (e.g., NVIDIA Nsight, AMD Prof, Intel VTune).
Responsibilities
Develop and optimize HPC software solutions leveraging NVIDIA GPUs and AMD CPUs for maximum efficiency.
Employ parallel computing techniques (MPI, CUDA, OpenMP) to ensure effective application scaling on multi-core and multi-GPU systems.
Contribute to the design and scaling of HPC nodes to meet computational demands, optimizing software and configurations.
Identify and implement algorithms suited for GPU/CPU computing to handle complex, data-intensive tasks.
Conduct profiling and benchmarking of HPC applications to identify performance bottlenecks and ensure consistency across nodes.
Collaborate with scientists, engineers, and IT staff to optimize HPC applications for various environments.
Create clear documentation for code and scaling processes, reporting performance metrics and improvements to stakeholders.
Benefits
Opportunity to work with cutting-edge technology in high-performance computing.
Collaborative work environment with cross-functional teams.
Access to advanced tools and frameworks for performance optimization.
Potential for professional development and further education opportunities.
Flexible work arrangements and a focus on work-life balance.
Full Job Description
Job Summary:
As an HPC Software Engineer, you will be responsible for designing, developing, and optimizing HPC software solutions that leverage NVIDIA GPUs, AMD CPUs, and MPI for distributed and parallel computing. This role will require a strong foundation in computational science, parallel algorithms, and the specific intricacies of GPU and CPU optimization. You'll also play a key role in designing and implementing scalable solutions for our HPC clusters, ensuring high-performance applications can run efficiently as nodes scale up to meet expanding global workloads.
Responsibilities:
Develop and Optimize HPC Software: Design and implement software solutions that harness the power of NVIDIA GPUs and AMD CPUs, focusing on maximizing computational efficiency and speed.
Parallel and Concurrent Computing: Employ expertise in MPI, CUDA, OpenMP, and other parallel computing techniques to ensure that applications scale effectively on multi-core and multi-GPU systems.
Scaling and Cluster Management: Contribute to the design, deployment, and scaling of HPC nodes to ensure clusters can expand based on data and computational demands. Optimize software and cluster configurations to enable seamless scaling of nodes.
Algorithm Optimization: Identify and implement algorithms best suited for GPU/CPU computing to handle complex computations and data-intensive tasks, with a focus on scalable solutions that maintain performance as node count increases.
Node Profiling and Performance Analysis: Conduct profiling and benchmarking of HPC applications across scaled environments to identify bottlenecks and ensure consistent performance across nodes. Use tools such as NVIDIA Nsight, AMD Prof, and Intel VTune for performance tuning.
Collaboration Across Teams: Work closely with scientists, engineers, and IT staff to ensure HPC applications are optimized and scalable for various research and production environments.
Documentation and Reporting: Create clear documentation for all code and scaling processes. Report on performance metrics, improvements, and results to stakeholders.
Experience/Qualifications:
Education: Bachelor's or master's degree in computer science, Electrical Engineering, Computational Science, or a related field. Advanced degrees or relevant certifications are a plus.
Experience: 3+ years of experience in HPC development, with a focus on parallel and distributed computing, and experience with scalable systems. 5+ years previous experience in scientific software development.
Technical Skills: NVIDIA GPU Programming: Proficiency in CUDA and familiarity with NVIDIA's HPC software stack, including cuDNN, NCCL, and TensorRT.
Parallel Programming: Expertise in MPI, OpenMP, and other parallel programming paradigms.
Cluster Scaling and Optimization: Strong understanding of scaling clusters and optimizing software for distributed, multi-node environments.
Languages: Strong proficiency in C/C++, Python, and Ansible / Terraform / Packer
Tools and Frameworks: Familiarity with profiling and performance optimization tools (e.g., NVIDIA Nsight, AMD Prof, Intel VTune).
Soft Skills: Ability to work independently and as part of a team, excellent problem-solving abilities, strong communication skills, and a keen attention to detail.
Preferred Skills
Familiarity with HPC Environment Management: Experience with Slurm, PBS, or other job scheduling and resource management tools.
Experience with Containerization: Knowledge of Singularity or Docker in an HPC context.
Experience with Cloud Scaling Solutions: Understanding of cloud-based HPC scaling solutions or hybrid HPC environments is a plus.
Machine Learning Acceleration: Experience with machine learning frameworks that utilize GPUs, such as TensorFlow, PyTorch, or PySCF is a plus.
About PsiQuantum
PsiQuantum is a California-based company that is developing a quantum computer using photonics technology. The company's goal is to build a fault-tolerant quantum computer with millions of qubits, which would be capable of solving problems that are currently intractable with classical computers. PsiQuantum was founded in 2016 by a team of experts in photonics, quantum computing, and computer science. The company has received funding from investors such as BlackRock, Founders Fund, and Atomico.
