This project introduces an automated audit report generator designed to assist financial institutions in producing high-quality, regulator-ready audit reports. Auditors often work under tight deadlines while navigating hundreds of pages of heterogeneous documents, creating inefficiencies, inconsistencies, and onboarding challenges for new team members. Our solution uses a RAG-based architecture to ingest PDFs, semantically chunk and vectorize content, retrieve the most relevant information, and generate a comprehensive audit report grounded in real documents. The prototype focuses on ICAAP credit risk — a demanding, document-heavy domain — using Streamlit for the interface, FastEmbed for lightweight embeddings, and Qdrant as the vector store. The system reduces manual workload, enhances audit consistency, strengthens internal risk-management controls, and ultimately helps institutions mitigate regulatory risk and avoid supervisory findings.
Category tags: