Performance Optimization - Postdoctoral Researcher

Key Skills

C Programming, C++, Effective communication skills, Linux Operating system, Machine learning techniques, Python Programming, PyTorch, TensorFlow, Unix Operating system

Job Description

We have an opening for a Postdoctoral Researcher to perform research in performance optimization of the software stack used in High Performance Computing (HPC). You will work collaboratively with LLNL researchers to research and develop new methodologies to optimize the execution of scientific applications targeting state-of-the-art heterogeneous many-core HPC systems. This position is in the Center for Applied Scientific Computing (CASC) Division within the Computing Directorate.

You will 

• ​​​​​​Research, design and implement scalable tools combining static and dynamic techniques to identify application performance issues.
• Research, design and develop mechanisms to correct possible performance issues in application codes or in HPC programming models and runtime libraries.
• Research and develop software representations suitable for extracting performance characteristics using analytical or machine learning methods.
• Explore novel schemes to achieve scalability in detecting and correcting performance issues.
• Document research by publishing papers in peer-reviewed media and presenting papers within the DOE community and at academic conferences.
• Contribute to group grant proposals, including proposal presentations and preparation of proposals that will provide future research opportunities in the field, and participate in the establishment of future research directions.
• Pursue independent (but complementary) research interests and interact with a broad spectrum of scientists internally and externally to define and carry out the research.
• Perform other duties as assigned.

Qualifications

• Recent Ph.D. in Computer Science or a related field.
• Experience in designing and implementing parallel computing algorithms.
• Ability to perform research and development in compiler optimizations or HPC programming models, such as OpenMP, RAJA, and CUDA.
• Ability to conduct high quality independent research and to develop implementations to evaluate the results.
• Experience in programming C/C++ in a Unix/Linux environment.
• Proficient verbal and written communication skills necessary to interact in a clear and concise manner, author technical and scientific reports and papers, and deliver scientific presentations.
• Ability to take the initiative and have interpersonal communication skills necessary to work effectively in a dynamic team environment.

Qualifications We Desire

• Experience in compiler or programming model runtime development, preferably in the Clang/LLVM OpenMP compilation/runtime framework.
• Experience in parallel programming, preferably in widely used parallel programing models, such as OpenMP, CUDA, HIP and/or MPI.
• Knowledge of machine learning frameworks, such as PyTorch or Tensoflow, and other scripting languages, such as Python or Lua.