Fine-tuning an LLM to extract insights from UNDP Project Documents. This thesis is set out to explore the possibility of using LLMs to create a knowledge engine trained on the UNDP Project documents, which could be used variously to support theUNDP. The immediate application of interest it to develop a model that can estimate direct and indirect beneficiaries from projects supporting energy access.  You can see the projects at the UNDP Transparency Portal here - https://open.undp.org/projects, and the project documents can be found in the "Documents" section of each project and are typically titled "Prodoc".

Work alongside the UNDP // Project 2 with the UNDP "Sustainability, indicators and machine learning: Case Study with the United Nations Development Program"