ROLE RESPONSIBILITIES
The successful candidate will become a go-to technical expert and key support contact for ML/AI researchers working in internal and cloud based High Performance Computing (HPC) environments.
Interface between computational scientists and HPC engineers responsible for maintaining hardware and other core services to maximize the benefit of HPC and cloud capabilities. Provide guidance to IT operational teams and promote HPC/Cloud adoption within the community.
Install, compile, troubleshoot, optimize, and document use of scientific software in a scheduled Linux computing environment. Includes building, updating and troubleshooting Python environments, containers, commercial, and in-house applications.
Develop user documentation and training targeted at application scientists for scientific computing hardware and software.
Attend technical meetings and relevant scientific conferences.
BASIC QUALIFICATIONS
Advanced degree in a relevant field or 2-4 years of experience in ML/AI, life sciences, data science, or HPC
Excellent Linux skills, including scripting, building/installing software, using package managers or managing scientific software in an HPC environment
Familiarity with HPC environments
Working knowledge of open source frameworks used to develop ML applications (PyTorch, Tensorflow, Keras, SciPy/NumPy, Pandas)
Demonstrated ability to troubleshoot issues related to machine configuration, environment settings, etc.
Excellent written and verbal communication skills; good in a team environment.
Demonstrated ability to juggle multiple tasks, dynamically determining priority based on need, while keeping all customers apprised of the status of their request(s)
PREFERRED QUALIFICATIONS
Ph.D. in data science, life science, computer science or other related field
Fluency in Python
Advanced HPC experience
Experience implementing new ML/AI workflows
The candidate will work in HPC and cloud environments with tools such as Python, Jupyter, Bright, Ansible, AWS, Slurm, Docker, Singularity, Git, and Terraform. Familiarity with some (but not all) of these tools is expected.