• Used 3D CNN to analyze low-dose CT scan data and predict the likelihood of lung cancer. Achieved 64% accuracy in classifying patients as either suffering from cancer or not.

[Research Paper Published Wily Publication, M.Tech Major Project]

• Developed an efficient and reliable method using advanced CNN PyTorch models by Google AI, EfficientNet, to detect secret data hidden within digital images with a minimal false positive rate. Achieved an F1 score of 77.7%.

[M.Tech Research Thesis]

• Goal was to classify fraudulent and non-fraudulent credit card transactions. The accuracy was measured using the Area under the Precision-Recall Curve (AUPRC) as confusion matrix accuracy is not meaningful for unbalanced classification. Achieved Model Accuracy of 80%.

[B.Tech Minor Project]

• The project aimed to generate captions for images using a pretrained image-model (VGG16) to create a "thought-vector" representing the content of the image. A Recurrent Neural Network (RNN) was then trained to map this "thought-vector" to a sequence of words, achieving a model accuracy of 60%.

[Internship Project]

• This project provides real-time data updates every 5 minutes on popular global topics or specific keywords. It displays relevant tweets in real-time on the console using Scala and the Spark Streaming library for efficient data processing.

[Internship Project]