Context Aware Token Optimization for Personalized Financial Advisory with LLMs

Sarkar, Biswadeep (2025) Context Aware Token Optimization for Personalized Financial Advisory with LLMs. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

Generative AI has transformed both industry and academia. The extensive ability of these models to process, analyze, and generate context-aware responses has opened new horizons of possibilities across domains. In this paper, I have explored how Large Language Models (LLMs) can be used as an extended brain for financial advisors, enabling them to generate context-aware responses utilizing information previously shared by customers. While doing so, my goal was to optimize the input token count significantly. Following the work of previous researchers and after several iterations, I have created a multi-LLM financial advisor architecture powered by a multi-layer context extraction engine that generates context-aware responses with more than 92 percent reduction in input token count. The architecture employs both zero-shot and one-shot prompting techniques to balance generalization capabilities with contextual guidance. The system architecture is demonstrated through a Flask-based web application that mimics a WhatsApp-style chat interface using HTML, CSS, and JavaScript, with the backend implemented in Python. This efficiency translates to substantial savings in computational resources, energy consumption, and carbon emissions, highlighting the significant sustainability potential of the approach.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Byrne, Brian UNSPECIFIED
Subjects:	H Social Sciences > HG Finance P Language and Literature > P Philology. Linguistics > Computational linguistics. Natural language processing H Social Sciences > HG Finance > Fintech T Technology > T Technology (General) > Information Technology > Fintech Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence > Generative artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence > Generative artificial intelligence
Divisions:	School of Computing > Master of Science in FinTech
Depositing User:	Ciara O'Brien
Date Deposited:	20 Aug 2026 09:25
Last Modified:	20 Aug 2026 09:25
URI:	https://norma.ncirl.ie/id/eprint/9559

Actions (login required)

View Item