RolinaVorster0101/SALabourLaw

RAG-based South African employment law assistant built with ASP.NET Core, Azure OpenAI, and Azure AI Search. Supports legislation Q&A (BCEA, LRA, CCMA) and employment contract compliance review. # SA Labour Law & Contract Compliance Assistant ## Problem Statement South African employment law is dense, frequently misunderstood, and rarely consulted until there's already a dispute. The Basic Conditions of Employment Act (BCEA), the Labour Relations Act (LRA), and CCMA guidance govern everything from notice periods to leave entitlements to dismissal procedures, but most employers and employees never read the legislation directly — they rely on guesswork, outdated templates, or expensive legal consultations for questions that often have a clear, documented answer. This project applies Retrieval-Augmented Generation (RAG) to that problem. RAG is the architectural pattern behind every serious enterprise AI application in 2026 — "chat with your documents," internal knowledge bases, document Q&A systems — and it directly addresses the core limitation of LLMs: they only know what they were trained on. RAG feeds them fresh, specific, authoritative information at query time before they generate an answer, rather than letting the model guess from general training knowledge. The tool operates in two modes. A general Q&A mode lets any user ask questions directly against the BCEA, LRA, and CCMA guidance — "how much notice is required to terminate an employee," "what are the rules around Sunday pay." A contract review mode lets a user upload their own employment contract and ask whether specific clauses comply with the legislation, with the tool retrieving relevant sections from both the uploaded contract and the legislation simultaneously and flagging discrepancies with citations back to both sources. This project is both a practical tool and a direct lab for the AI-103 certification (Azure AI Apps and Agents Developer Associate), covering exactly the concepts that exam tests. ## Architecture Legislation ingestion pipeline: the BCEA, LRA, and CCMA guidance documents are chunked into overlapping segments, each chunk is converted into a vector embedding using Azure O …