Ankit Binu

• Developed dynamic neural networks to optimize computational efficiency by reducing FLOPs while maintaining model accuracy.
• Acquired hands-on experience in model optimization techniques for deployment in resource-constrained environments.

• Developed a classification model using machine learning to classify medical images of human cells.
• Built a classification model using customer data to forecast insurance purchases, from data preprocessing to final model creation.

TRIPMATE: AI-Powered Travel Buddy & Recommender

Emotion-aware tour recommender with smart booking redirection. Delivers personalized destination suggestions using user preferences, travel history, sentiment analysis, and multimodal emotion detection (facial emotion recognition using CNN + questionnaire mood analysis). Includes collaborative filtering for buddy matching.

Dynamic Layer Skipping for Neural Networks

Implemented a neural network that dynamically skips layers using gating mechanisms to significantly improve inference speed with minimal accuracy loss for deployment in resource-constrained environments.

Breast Cancer Classification Model

Developed and trained a Logistic Regression classification model to accurately analyze histology medical images from the public IDC dataset.

Customer Insurance Purchases Predictive Model

Built an AI model to predict whether new customers will purchase insurance using demographic features such as age and estimated salary.