Paper Title
Hybrid Graph-Augmented Retrieval for Intelligent Developer Documentation Question Answering

Abstract
Developers commonly face difficulties in locating specific details within large - scale documentation repositories such as technical handbooks, code documentation, project summaries and READMEs. Existing search engines typically rely on keywords and lack a deep understanding of content context. Consequently, these systems often produce irrelevant results for developer queries. DevDocs is introduced as a sophisticated AI helper that can answer developers' questions on code, API references and projects by means of retrieval - augmented generation (RAG) with vector embeddings and similarity search. A developer may upload their documentation, after which the program generates embedding for the entire set of documents, then saves the embeddings and source documents into a vector database. Afterward, when developers pose questions in plain natural language, DevDocs will conduct a similarity search against relevant documents, then present a limited set of context - rich results along with their original source references and use the responses into a generative AI system (LLM) that answers questions. Thus, this system brings the benefits of both semantic search, indexing - enabled structured indexing and chatbots in software documentation, offering increased developer efficiency, reduced information - finding time, scalability and manageability. Keywords - Retrieval-Augmented Generation, Developer Documentation Systems, Semantic Search, Vector Databases, Large Language Models, AI-Powered Documentation Assistants, Knowledge Retrieval Systems