Rohan Bapat

• Designed and implemented an email notification system to track the upload of profit data from financial teams, send monthly reminders to upload new data, and aggregate upload information in a new web dashboard to display metrics about delayed uploads.
• Email system and web dashboard was built to reduce the number of delayed uploads, keep product profitability data up to date, and boost accuracy of profit predictions in downstream teams.

• Developed an extensible framework to convert several different types of API specifications into a universal and simplified format, which was imported and used by several teams to automate their workflows.
• Integrated framework with existing development kit to provide build-time static validation of generated code.
• Drove project to completion, designing it to promote easy-adoption by other teams and refactoring it to integrate with the teams larger code-base.

Systems Engineering Intern

• Created a proof of concept for a real-time object detection pipeline using machine learning algorithms on concurrent camera streams.
• Frames were streamed from IP Cameras to a python back-end where a pretrained convolutional neural network was run.
• Detection pipeline was used to detect and draw a bounding box over common objects in a video stream to demonstrate the capabilities of a processing unit.

Research - Predicting the onset of Dementia

• Built a convolutional neural network with PyTorch to predict the onset of Dementia three years in the future with 85% accuracy, using only patient MRI scans and cognitive test scores.
• Current model outperforms other models from recent literature that use the ADNI dataset for this task.
• Currently working with Dr. Tim Duong on a publication for this research.

University of Massachusetts Amherst

University of Massachusetts Amherst