Language/Tools used: Python, Jupyter Notebook, AWS, sshLibraries used: Tensorflow, scikit-learn, RAPIDS, pandas, NumPy, matplotlib, seaborn, RDkitProject Summary: Development of Machine Learning Algorithms for Small Molecule Discovery and Chemical Space Analysis• Performed exploratory data analysis and visualization of chemical data to prepare data for model training.• Trained supervised machine learning algorithms such as Deep Neural Networks, Random Forest (RF) Classifiers/Regressors, Support Vector Machines (SVM) on AWS Amazon Machin Images(AMIs), and Teslav100 GPU to predict molecule activity.• Decreased training time and improved model performance by using Principal Component Analysis (PCA) for dimensionality reduction.• Integrated trained models in the reward systems of state-of-the-art Reinforcement Learning Algorithms to find potential novel tracers for PET Imaging.• Improved model performance by decreasing Root Mean Squared Error (RMSE) for regression models and increasing F1-scores for classification models through hyperparameter tuning via cross-validation methods like K-Fold and GridSearchCV. • Please Visit the GitHub repository ‘https://github.com/Shiska07/Cheminformatics-and-Drug-Discovery’ for project files and more information.

Project: Medical Image Segmentation Application for efficient 3D ReconstructionLanguage/Tools used: MATLAB Image processing toolbox, MATLAB App Designer• Built chest CT segmentation application with an interactive GUI by utilizing MATLAB App Designer and Image processing toolbox.• Reduced average total time for 3D reconstruction from 1 week to 48 hours through automated image segmentation via edge detection, histogram equalization, and binary masking.• Please visit GitHub repository ‘Shiska07/Medical-Image-Processing/MATLAB-Chest-CT-Segmentation-App’ for project files and additional information.

Introduction to Problem Solving Techniques in EngineeringSI leaders are responsible for conducting student-driven study group sessions for the most difficult sections of courses that are historically known to be challenging. SI leaders promote independent learning and facilitate students towards appropriate study resources on campus.

Peer tutor in STEM related courses.