Yeementh Virutkar

ESG Document Intelligence with GenAI

• Developed an ESG document analysis system using LLMs (Gemini Multimodal, Claude) to classify and evaluate company sustainability reports.
• Built pipelines to:
  Classify files and extract organizational identities.
  Categorize content into Environmental, Social, and Governance sections with subcategories.
  Determine ESG qualification based on domain-specific rule-based metrics.
• Implemented Retrieval-Augmented Generation (RAG) for real-time QA over ESG content using:
  LangChain and LlamaIndex for flexible orchestration.
  ChromaDB for vector storage and retrieval.
• Delivered a scalable, multi-model GenAI architecture for enterprise ESG compliance.

DOCIPHI – Intelligent Document Processing Platform

• Spearheaded GenAI integration for unstructured insurance documents using LLM agents for intelligent file classification and deep content extraction.
• Led development of a table extraction module for diverse insurance file formats with context-preserving accuracy.
• Designed and maintained Google Cloud-based pipelines to host and serve multimodal LLM models (Claude, Gemini), enabling real-time document understanding at scale.
• Enhanced document intelligence through modular LLM microservices integrated via API, laying groundwork for Agentic RAG and downstream compliance tools.