Mid-Atlantic HERC

LEAD RESEARCH SOFTWARE ENGINEER, PORTABLE AI PERFORMANCE ENGINEERING, MA Green High Performance Computing Center, to be a hands-on research software engineering professional and serve as lead for applied performance engineering for AI workloads. Will work closely with research groups and leading computer industry collaborators to evaluate, adapt, and enhance the portable performance of complex AI research workloads on state-of-the-art hardware. The role will have heavy focus on optimizing existing NVIDIA GPU-based workloads for top-tier AMD GPUs, such as MI355X and beyond and will analyze and profile existing research AI workloads to identify performance bottlenecks and portability challenges; and port and optimize complex AI models and scientific code to run efficiently on AMD MI355X GPUs using ROCm, HIP, and related translation tools.

REQUIRED: Bachelor's degree or equivalent with a minimum of five years of work experience in either deeply technical fields and/or computational research experience; strong proficiency in Python and C++, with deep familiarity with AI/ML frameworks (PyTorch, TensorFlow, JAX); hands-on experience with GPU programming models (e.g., CUDA, HIP, or OpenCL); experience with performance profiling and benchmarking tools on Linux-based High-Performance Computing systems; excellent communication skills; ability to collaborate effectively with academic researchers and industry partners; and self-motivated with the ability to work independently in a remote or hybrid environment. PREFERRED: Direct experience with the AMD ROCm software stack and translating CUDA code to HIP; familiarity with AI agentic tools and Large Language Models (LLMs) used for code generation and refactoring; background in supporting large-scale, domain-specific scientific research (e.g., physics, biology, climate science) on institutional clusters; direct experience with one or more open-source schedulers and provisioners; experience with Linux container technologies such as LXC, apptainer and systemd-nspawn; or advanced degree in a relevant technical field.

The Lead Software Engineer must comply with all relevant MGHPCC security policies.

This is a two-year term position.

3/13/2026