Pioneering AI Agent Orchestration for Enterprise Automation

Led by Senior AI/ML Engineer Mr.Vidyasagar Reddy Parlapalli, SAGE (Service AI Generating Expertise) revolutionizes technical support for customers like Intel / Samsung /TSMC / Micron through generative AI and agentic architectures, achieving a significant reduction in issue resolution times. This initiative deploys an enterprise-grade LLM chatbot that integrates multimodal engineering (text, tables, images), Retrieval-Augmented Generation (RAG), and autonomous AI agents to transform KLA’s customer experience.

The project pioneers an advanced RAG framework that combines query expansion, knowledge graphs, and self-query retrieval to optimize context-aware responses. By engineering dynamic chunking strategies and RAFT tuning, the system minimizes hallucinations while boosting retrieval precision by 40%. This innovative approach enables the LLM chatbot to handle complex queries more effectively, reducing the reliance on manual intervention.

At the core of SAGE is an agentic AI architecture designed with specialized roles: a Question Dissector, Hallucination Evaluator, and Solution Generator, all finetuned using QLORA/PEFT for parameter efficiency. This multi-agent system is integrated with TGI and vLLM for high-speed inference, allowing it to handle concurrent queries with sub-second latency. The MLOps pipeline ensures seamless model updates and scalability, automating CI/CD processes for continuous improvement.

By leveraging reinforcement learning from CSE feedback and email conversation mining, the model iteratively refines its performance, enhancing user satisfaction. The project has achieved 85% user satisfaction in pilot testing and is projected to scale across multiple domains. Notably, SAGE has saved Technical Support Engineers approximately 30% of their time by automating routine troubleshooting tasks, thereby increasing productivity and efficiency.

This project positions KLA at the forefront of AI-driven customer support, merging agentic reasoning, multimodal RAG, and self-improving systems to set a new industry benchmark. By building the foundation for KLA’s AI Center of Excellence, Mr.Parlapalli’s work exemplifies how autonomous AI can redefine human-machine collaboration, driving innovation and efficiency across the semiconductor industry and beyond.