Process Modeling using Physically Informed Machine Learning (CHIPS Funded Project)

Overview

The work will entail:

• Designing and training physics-informed machine learning (PIML) models for the prediction of physical and chemical properties using data from experiments and computation constrained by physics requirements.
• Implementing algorithms to assess the performance of PIML models.
• Assessing uncertainty in the predictions of PIML models.
• Developing systems for multiscale modeling of atomic layer deposition processes.
• Developing software to implement the goals stated above (most likely in Python).
• Disseminating results through posters/seminars and international meetings and meeting seminars.
• Ensuring that all results, findings, data, software, etc. are correctly archived and transmitted through appropriate channels.

Qualifications

• A Ph.D degree in Chemistry, Physics, Mathematics, Computer Science, Data Science, or a related field.
• Significant course work in one or more of chemistry, physics, mathematics, statistics and/or computer science.
• Familiarity with one or more chemical process modeling packages (e.g. Cantera, CHEMKIN).
• Familiarity with one or more AI/ML software packages (e.g. Tensorflow or Pytorch).
• Ability to program in a modern computational language (e.g. Python).
• Strong oral and written communication skills.

Research Proposal

Key responsibilities will include but are not limited to:

• Algorithm development, implementation, and analysis
• Analyze heterogeneous data sources.
• Presenting results at internal meetings, and occasional meetings with external stakeholders.
• Ensuring that results, protocols, software, and documentation have been archived or otherwise transmitted to the larger organization.