Sharan Shyamsundar

Contributed to the development of an internal helpdesk powered by Retrieval-Augmented Generation, reducing query resolution times by 20%. Built interactive dashboards for supply chain analysis and streamlined operations.

Researched Parameter-Efficient Fine-Tuning techniques for large language models, achieving significant memory reduction while maintaining accuracy. Evaluated efficiency trade-offs in model deployment.

Developed an AI-driven cooking assistant integrating ASR and LLMs, achieving high accuracy in recipe generation. Enhanced user experience with hands-free navigation and multilingual support.

Participated in the Informatics Cup, developing a solution for detecting AI-generated content. Implemented ensemble models for enhanced accuracy and addressed AI security challenges.

Led ESG-related financial analytics products, ensuring compliance with best practices. Developed advanced analytics solutions for hedge funds, enhancing decision-making processes.

Consulted on financial reporting and risk management, automating workflows and improving data accuracy. Spearheaded a database transition, reducing downtime and boosting processing speeds.

Key courses included Introduction to Data Science, Time Series & Forecasting, Multivariate Analysis, Hypothesis Testing, Estimation Theory, and Multivariate Calculus.

Key courses included Advanced Text Analytics, Deep Learning, and Machine Learning. My seminar focused on Large Scale Data Integration - Deep Tabular Learning for Domain-Specific Prediction Tasks. My thesis was a study of Parameter-Efficient Fine-Tuning Methods applied to Quantized Large Language Models.