Union Bank of India – Anand Das – 2025

Summary

Anand worked on a proof of concept for Union Bank of India to automate the classification of loan advances into climate change mitigation, adaptation, or non-climate finance by building a machine learning-based model.

Goals

Union Bank of India aimed to explore the possibility of automating the classification of its advances under climate finance. Union Bank of India enlisted the Climate Corps fellow, Anand Anubhav Das, to design a prototype model using Machine Learning for accurate classification of loan proposals.

Solutions

Anand conducted industry benchmarking and mapped requirements for this project. He built a Python-based model that used semantic analysis for classification, following conventions and guidelines from major organizations such as WB, MDB, ICMA, LMA, RBI, etc. The model integrated an LLM for natural language processing and RAG (Retrieval Augmented Generation) for guidance. The model was tested on scenarios of loan proposals in a standard format and publicly available TEV (Techno Economic Viability) project reports. The model worked satisfactorily on these scenarios, demonstrating its potential for further refinement and subsequent deployment in the bank’s internal software.

Potential Impact

If this project is refined, implemented, and deployed by the bank, it could help the bank to analyze its climate finance data, reduce resources and manpower needs, and increase its ESG scores from international rating agencies.