Smit Chaudhari

Enthusiastic Computer Science graduate with a strong academic foundation and hands-on experience in machine learning, data analytics, and software development. Proven ability to develop scalable ML pipelines, analyze large datasets, and communicate results with clarity. Skilled in Python, SQL, scikit-learn, and data visualization. Eager to contribute analytical thinking and technical skills to data-driven roles that solve real-world problems.

• Deep Learning Pipeline – Galaxy Morphology Classification (Dissertation, 2025):
• Built an end to end ML pipeline using PyTorch to classify 145,000+ galaxy images from the Galaxy Zoo 2 dataset.
• Achieved 64% accuracy and 0.51 macro F1-score using fine-tuned ResNet-18 with transfer learning.
• Implemented Grad-CAM++ for interpretability, visualizing model attention areas.
• Happy vs Sad Image Classification (Coursework, 2024):
• Trained Random Forest using CNN & GIST features, with 74% cross-validation accuracy.
• Applied robust pre processing, imputation, and feature importance analysis.
• Sentiment Classification (NLP) (2023):
• Developed pipeline using Naive Bayes, SVM, and Logistic Regression.
• Handled tokenization, stopword removal, and stemming using scikit-learn.

Volleyball, Fitness & Gym, Coding side projects, Reading web novels, Music