Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{178140,
        author = {Chaithali SK and Janhavi Sachdev and Prof. Kavita Babalad},
        title = {Beyin:An Integrated Framework for Multi-Expert LLM Orchestration with RL-Based Routing},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {12},
        pages = {5874-5883},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=178140},
        abstract = {Project Beyin presents a novel framework that integrates multiple domain-specific Large Language Models (LLMs) with an intelligent routing mechanism. This paper introduces a reinforcement learning-based dynamic query routing system designed to address subject-specific queries with high precision. We implement Parameter-Efficient Fine-Tuning (PEFT) techniques, specifically Low-Rank Adaptation (LoRA), alongside sophisticated retrieval mechanisms to deliver factually grounded, context-aware responses. Our methodology encompasses three key components: (1) fine-tuning multiple LLMs for targeted domains using specialized datasets, (2) implementing an advanced Retrieval-Augmented Generation (RAG) pipeline with embedding-based retrieval and contextual re-ranking, and (3) developing a Proximal Policy Optimization (PPO) based routing system that dynamically directs queries to the appropriate domain expert model. Experimental results demonstrate that Beyin effectively reduces hallucinations, enhances contextual relevance, and scales efficiently across diverse applications including healthcare, education, and enterprise knowledge management. This research addresses critical gaps in existing frameworks by providing an integrated solution for domain-specific AI challenges. Future work will focus on incorporating real-time feedback mechanisms and expanding multimodal integration capabilities.},
        keywords = {Retrieval Augmented Generation (RAG), AI Agents, Large Language Models (LLM), Knowledge Graphs (KG)},
        month = {May},
        }