Fragan Dsouza

About Fragan Dsouza

Computer Science undergrad.Building myself like good software - modular, versatile, and finely tuned for excellence.Working to be a jack of all trades, mastering one at a time.

Experience

1. Artificial Intelligence Intern

   Inspirante Technologies Pvt. Ltd

   Feb 2026 — Present

   Key responsibilities:• Designing and optimizing deep learning architectures, including ResNet50, DenseNet121, and MobileNetV2, for automated classification of brain MRI and CT scans.• Implementing Grad-CAM (Gradient-weighted Class Activation Mapping) to enhance model interpretability by visualizing the regions of interest influencing diagnostic predictions.• Conducting comparative performance benchmarking across multiple deep learning models using metrics such as Accuracy, AUC-ROC, and computational efficiency for clinical deployment.• Developing automated ROI extraction pipelines that generate localized bounding boxes for detected lesions, enabling faster radiological review.• Performing data engineering and dataset balancing on large-scale medical imaging datasets to mitigate class imbalance and improve model generalization.• Applying advanced regularization techniques such as Early Stopping and Dropout to improve training stability and prevent overfitting on sensitive medical datasets.• Collaborating on system architecture design, research documentation, and reporting of diagnostic performance metrics for clinical validation.

Education

• Indian School Muscat

  Computer Science

• NMAM Institute of Technology

  Bachelor of Technology - BTech, Computer Science