Shubhankar Tripathy

Bachelor's

Mathematics and Computer Science

Personalized Class Notes Summarization System

- Python, QLoRA, Transformers, Hugging Face, PyTorch - System Design: Developed a system to summarize unstructured lecture data into concise, personalized - Scalability: Designed for horizontal scaling using AWS Lambda and S3 for efficient cloud deployment - Resource Optimization: Applied QLoRA to reduce resource consumption by 50% while maintaining model - Efficient Preprocessing: Built a pipeline for tokenization, stopword removal, and key concept extraction - Model Accuracy: Achieved 88% accuracy in generating context-aware summaries using ROUGE and BLEU - Impact: Improved study efficiency by enabling students to review personalized summaries