DeepTutor is an open-source multi-agent framework for automated tutoring and research assistance that utilizes large language models (LLMs) to facilitate interactive learning and idea generation.

• Integrates Retrieval-Augmented Generation (RAG) with dynamic knowledge graphs to provide factually grounded educational content and structured data retrieval.
• Employs a multi-agent architecture to orchestrate specialized workflows for information retrieval, data synthesis, and pedagogical interaction.
• Automates deep research tasks by performing structured literature analysis and generating technical hypotheses based on user-defined parameters.
• Supports stateful, interactive learning sessions through agentic prompting and semantic search capabilities across diverse datasets.

Built on Python, the system enables the integration of various LLM backends to optimize high-dimensional data processing and pedagogical accuracy. This framework is designed for academic research automation, personalized STEM tutoring, and enterprise knowledge synthesis.