Atul Prem Patnala

Research Assistant — University of North Carolina Wilmington (UNCW) Under the supervision of Dr. Karl Ricanek As a Research Assistant, I contributed to applied machine learning and data science research in the areas of biometrics, driver identification, and temporal data modeling. My work involved: Designing and implementing data preprocessing pipelines for high-dimensional, time-series sensor datasets (CAN-bus telemetry). Developing and evaluating multiple machine learning models (Decision Trees, Random Forests, KNN, SVM, and Neural Networks), with a focus on window-based classification and feature importance analysis. Applying data segmentation, sliding window techniques, and stratified evaluation methods to ensure robustness and avoid data leakage. Conducting comparative performance analysis across traditional ML and deep learning models, reporting findings via classification reports, confusion matrices, and feature interpretability tools. Supporting the team with experimental documentation, reproducibility, and model interpretation for advancing ongoing projects in driver profiling and behavioral biometrics. This role strengthened my expertise in Python, scikit-learn, TensorFlow/Keras, feature engineering, and model interpretability, while fostering collaboration in an academic research setting.

Teaching Assistant | CSC 322: Introduction to Data & Machine Learning Instructor: Dr. Gulustan Dogan University of North Carolina Wilmington (UNCW) As a Teaching Assistant for CSC 322, I support students in understanding key concepts in machine learning and data science. My responsibilities include assisting with lab sessions, grading assignments, and providing guidance on topics such as data preprocessing, classification, predictive modeling, and big data analytics. Technologies Used: Python, Scikit-Learn, Pandas, NumPy, Matplotlib etc.

Project: Visualization of Thermal Flow Patterns on Control Surfaces of Factory Apparatus Developed a software application to analyze thermal imaging data for temperature mapping. Implemented image processing techniques to extract and collate thermal data from multiple images. Designed an automated data visualization tool to generate a comprehensive thermal map of factory apparatus surfaces. Enhanced temperature pattern analysis by stitching thermal data from different imaging perspectives. Applied data analytics and visualization to improve decision-making in industrial thermal diagnostics.