Gaurav Shetty Email & Phone Number

Danville, Pa., Us

- Leading a team of 5 to develop and operationalize a CNN-based model for identification of intracranial hemorrhage using head CT scans which acts as a prioritization tool for identifying high-risk patients in clinical workflow. - Leading the technical team of 8 to develop the self-hosted Geisinger Internal GPT which is HIPAA compliant and designing the RAG architecture incorporating Geisinger’s Internal Knowledge base. - Developed a risk assessment model for Long COVID and hypothesized a clinical workflow integration to deliver patient care as a part of team project; the XGBoost model generates a risk score for Long COVID using a broad range of clinical data available in OMOP format. [Won the second place in NIH-sponsored N3C Long COVID challenge and $150,000 award] - Generated multi-modal deep learning-based models for risk stratification of patients to direct and develop appropriate palliative care, at-home care, and hospitalization strategies using electronic health records and screening images. [Presented at the AMIA Symposium 2022, 2024]. - Conducted in-department tutorials for computer vision and image processing.

Ann Arbor, Mi, Us

As a part of the Computational Medicine & Bioinformatics: Improved a pre-built deep learning-based survival prediction model (DeepProg) to improve results using different modalities of medical data - genomics data and histopathology image data; improvements of 13% were recorded for the c-index using the modifications.

Buffalo, Ny, Us

Buffalo, New York, Us

• Formulated a novel data modeling framework for data reconstruction/unsupervised regression by exploiting the underlying low-dimensional manifold employing manifold learning, dimension reduction, matrix factorization schemes.• Applied this novel framework to generate high resolution, high fidelity dynamic medical resonance images (dMRI) by acquiring fewer data from the respective sensors thus speeding up the scanning process in that pursuit.• Improved on the results for reconstruction of dMRI data by the introduction of kernel schemes into the data modeling framework.• Implemented the complete algorithm pipeline in MATLAB, Julia from data ingestion to the optimization framework to performing quantitative, qualitative analysis on the generated images and applied concepts from high-performance computing to generate time-efficient, parallelized codes.• Implemented various state-of-the-art CNN, GAN, and autoencoders based reconstruction schemes to study the learned representations in deep learning schemes and compare them against my developed frameworks.

Buffalo, New York, Us

- Draft assignment solutions, graded assignments- Conducted tutoring sessions to help students out with the coursework twice a week throughout the semester

Buffalo, Ny, Us

- Explored trends, correlations, any underlying information in the sensor & event data to determine potential damage signatures possibly leading to faults to thermal deterioration in generators using Python statistical and visualization libraries.- Employed an anomaly detection algorithm to detect trends that can predict damage progressions & created a thermal damage accumulation model to predict the current health state of a wind turbine.

Cambridge, Ma, Us

- Researched and investigated inter-dependencies between the Wikipedia usage patterns and the stock prices to further implement a predictor based on the research to validate the hypothesis using Python statistical and visualization libraries.- Applied a supervised learning scheme to predict stock market transitions utilizing the large, complex, high dimensional financial data sets on Yahoo Finance and view/edit logs by Wikipedia and using libraries like TensorFlow.- Gathered financial, page view, page edit data from Yahoo Finance, and Wikipedia using Python web scraping tools and PySpark.

- Tutored and mentored students from India, the United States, and the United Kingdom in Calculus and Statistics.- Influenced the initiation of the online teaching program (INK) and charged with the management of the same.- Generated Analytic Reports for preparing marketing schemes & strategies and responsible for data/record storage & management.

Ph.D., Electrical And Electronics Engineering

Master’S, Electrical And Electronics Engineering

Bachelor’S, Electronics And Telecommunications