Postdoctoral Appointee - HPC and Machine Learning
Job posting number: #7084353
Posted: August 31, 2021
Application Deadline: Open Until Filled
Job DescriptionWe are seeking self-motivated and independent postdoc researchers with strong background on Machine Learning, system design, and coding skills (C/C++). The selected candidate will work closely with the SZ compression team at Argonne to develop the cutting-edge lossy compression libraries/tools practically for the scientific community. The SZ team (http://szcompressor.org) is the leading team in the error-bounded lossy compression domain. The flagship software - SZ has been verified as one of the best error-bonded lossy compressors in the community by many domain scientists independently. Joining SZ team will have exceptional opportunities to collaborate with top-tier scientists in different domains and use cutting-edge supercomputers (Aurora, Summit, etc.).
Today's scientific applications are producing extremely large volumes of data, which are causing serious issues including storage burden, I/O bottlenecks, communication bottlenecks, and insufficient memory. Error-controlled lossy compression has been recognized as one of the most efficient solutions to resolve the big scientific data issue. Existing state-of-the-art lossy compressors, however, are all developed based on fixed/static compression models or pipeline, which cannot adapt to diverse data characteristics and sophisticated user-requirements. In this project, we will develop a scalable dynamic data reduction framework, which can optimize lossy compression for various use-cases dynamically and efficiently. The key techniques include using ML/DL to explore the diverse correlations of high-dimensional science datasets, using ML/DL to identify the best-qualified data compression model dynamically, using ML/DL to optimize the parameter configurations of various compression methods, using ML/DL to denoise datasets and/or recovering missing features for reconstructed data.
The selected candidate will work closely with the SZ compression team at Argonne to develop the cutting-edge lossy compression libraries/tools practically for the scientific community. Joining the SZ team will have exceptional opportunities to collaborate with top-tier scientists in different domains and use cutting-edge supercomputers (Aurora, Summit, etc.).
PhD degree in computer science, data analytics, or a related discipline.
Familiarity with machine learning to a certain extent. (being familiar with deep learning will be a plus)
Strong code development skills with C/C++. (being proficient in Python or Java will be a plus)
Familiarity with parallel environment (openMP, MPI). (being familiar with GPU will be a plus)
Strong publication record.
Strong skill in written and oral communications.
Good mathematics background
Experience in development of large-scale parallel systems
Familiarity with various data compression techniques
This job description documents the general nature of work but is not intended to be a comprehensive list of all activities, duties and responsibilities required of job incumbent. Consequently, job incumbent may be required to perform other duties as assigned.
Argonne is an equal opportunity employer, and we value diversity in our workforce. As an equal employment opportunity and affirmative action employer, Argonne National Laboratory is committed to a diverse and inclusive workplace that fosters collaborative scientific discovery and innovation. In support of this commitment, Argonne prohibits discrimination or harassment based on an individual's age, ancestry, citizenship status, color, disability, gender, gender identity, genetic information, marital status, national origin, pregnancy, race, religion, sexual orientation, veteran status or any other characteristic protected by law.