Research Scientist: ML for Collision Cross Section Prediction

Georgia Institute of Technology; Seattle (WA); On-site; USD 85,000 - 115,000; Full time

Job summary

The School of Chemistry & Biochemistry at the Georgia Institute of Technology invites applications for a Research Scientist I to advance the CCSP collision cross section prediction platform. The role focuses on developing, validating, and extending machine learning models that predict ion mobility values from molecular structure.

Responsibilities include training gradient-boosted and graph-based models, curating large MS datasets, and contributing to the public release of predictive tools used

Qualifications

• Demonstrated experience with machine learning in Python, including gradient boosting frameworks such as XGBoost and graph-based model architectures.
• Familiarity with cheminformatics tools and molecular descriptor generation, including RDKit and Mordred.
• Working knowledge of ion mobility spectrometry, mass spectrometry, and collision cross section measurement.
• Ability to communicate results clearly in writing and in oral presentations.

Responsibilities

• Develop, train, and validate machine learning models, including gradient-boosted decision trees and graph-based architectures, to predict collision cross section values from molecular structure.
• Curate, clean, and de-duplicate large collision cross section datasets, including the METLIN-CCS database, and generate molecular descriptors using RDKit and Mordred.
• Design and execute rigorous benchmarking studies, including multi-seed reproducibility analysis and fair comparison against existing predictors retrained on common data splits.
• Diagnose and resolve model performance issues such as overfitting and residual structure, and implement appropriate regularization and uncertainty quantification.
• Extend the prediction framework to additional compound classes, including lipids, and contribute to documentation and public release of the platform.
• Prepare figures, draft manuscripts, and present results at group meetings and scientific conferences.
• Contribute to a collaborative research environment and help mentor graduate and undergraduate students.

Skills

Machine learning

Python programming

Gradient boosting

Graph-based models

Ion mobility knowledge

Education

Bachelor's Degree in Chemistry or related field

Tools

RDKit

Mordred

XGBoost

Job description

The School of Chemistry & Biochemistry at the Georgia Institute of Technology invites applications for a Research Scientist I to advance the CCSP collision cross section prediction platform. The role focuses on developing, validating, and extending machine learning models that predict ion mobility values from molecular structure.

Responsibilities include training gradient-boosted and graph-based models, curating large MS datasets, and contributing to the public release of predictive tools used