● Collected and analyzed RehaGait inertial sensor data to extract spatiotemporal gait metrics and identify walking patterns. ● Developed MATLAB scripts to automate data quality assessment using statistical feature extraction. ● Automated data processing pipelines and managed Linux-based virtual machines for research projects. ● Delivered interpretable clinical results using Power BI to support decision-making. ● Administered collaboration platforms (SharePoint, Teams) to support interdisciplinary teamwork and project management.

Thesis: Classification of Hand Motion Patterns in Inflammatory Arthritis Patients using Machine Learning. ● Developed a high-dimensional time series classifier distinguishing arthritis subtypes with 86.2% accuracy using ensemble methods. ● Engineered features combining time-domain metrics and DCT transforms via NumPy/SciPy, processed through Scikit-learn pipelines. ● Accelerated model training by parallelizing Random Forest/XGBoost/SVM on HPC clusters (Slurm) using Linux-based systems. ● Validated performance against NIH diagnostic benchmarks using 5-fold cross-validation and Bayesian hyperparameter optimization. ● Boosted model precision by 12% using confusion matrix and feature importance, with strong evaluation of results.

● Designed and maintained dynamic dashboards and reports, facilitating data-driven decision-making for cross-functional teams. ● Documented pipeline architectures for sales analytics, enabling team maintenance. ● Developed dynamic sales and marketing reports with BI and KPI tools.

Machine Learning: Biomarker for Amyotrophic Lateral Sclerosis (ALS) Disease. (Biomedical Network Science LAB - FAU Erlangen). ● Implemented C++ modules using Gurobi for ALS/healthy classification on biomarker data, with version control via git and collaborative code reviews. ● Engineered features via integer programming and particle swarm optimization for robust selection. ● Validated models using confusion matrices for interpretable clinical results. Deep Learning: Robotics Arm with AI Vision for Dental Tool Handling. (Gujarat Technology University). ● Built CNNs for disease detection using DICOM medical images, supporting AI radiology research and clinical standards. ● Automated preprocessing and augmentation with OpenCV and DICOM tools for robust medical image analysis. ● Leveraged pre-trained architectures (VGG16, VGG19, AlexNet, ResNet) for image classification, achieving over 85% accuracy in disease detection. ● Collaborated with interdisciplinary teams to integrate results into clinical workflows, supporting DICOM and FHIR standards for data modeling and reporting.