Merck - Purdue Data Mine

(Machine Learning and Computer Vision Projects)

2022_01_29_Demo_SS_V2.pdf

Automation of Sample Management – As a part of the Machine Learning team I have developed a module to automate data retrieval from Material Safety Data Sheets (MSDS) using Text Analytics. Additionally, I have developed a code to merge multiple MSDS required for a given experiment and I created a warning system to alert MRL scientists about MSDS revision. During this process, I consistently updated and improved the code from the feedback I received from Merck’s scientist. This module when implemented will significantly reduce the time spent on manual entry of MSDS data at MRL.

Mentor: Dr. Terri Bui (Merck & Co.)


Automatic Documentation for Pharmaceutical Formulation – We are building an automated video narration framework that can be used to document the pharmaceutical production process. Our objective is to simulate and monitor laboratory experiments with cameras, and train Machine Learning models for object tracking and events and thereby automate the documentation. I have implemented a live QR code detection algorithm using Computer Vision techniques, developed a live motion detector for activity recognition in a laboratory setting, and built an OCR detector from live camera feed to extract product information from sample bags. This framework once completed and deployed can significantly reduce the time required for pharmaceutical formulation documentation.

Mentor: Dr. Ti-chiun Chang (Merck & Co.)