Vishva Patel

Master's

Computer Science

Cyber Security Systems

Anblicks

Machine Learning Research Intern

• Developed a CNN-based pneumonia detection system, achieving 89.1% accuracy on 5,800+ chest X-ray images • Diagnosed performance bottlenecks caused by data imbalance and suboptimal tuning, improving accuracy by 6.5% through preprocessing and model refinement • Assisted in integrating Vision Transformers (ViT) to address feature extraction limitations, boosting accuracy to 93.6% • Optimized GPU-based training pipelines, reducing model convergence time by 18% • Applied systematic hyperparameter tuning and evaluation strategies to improve model robustness and generalization • Developed a React-based web application with secure backend and frontend integration • Communicated findings to cross-functional teams, supporting informed technical and deployment decisions