Padaga Sravya
Kirthana

Architected an Agentic AI layer over HRMS for 10,000+ employees — specialized AI agents for HR Q&A, leave, payroll, recruitment, and analytics using RAG, NLP, tool calling, RBAC, and multi-agent orchestration with FastAPI backend.

Built an enterprise RAG platform enabling claims adjusters to query large volumes of claims documents, policy wordings, and process manuals in natural language using Azure OpenAI + GPT-4o with LLM-as-a-Judge evaluation.

NLP-based intent classification system to automatically triage high volumes of insurance emails (FNOL, billing, policy servicing, complaints) and route them to the correct operations queue. Fine-tuned BERT and RoBERTa with Sentence Transformers, deployed as FastAPI microservice in Docker with real-time and batch inference.

3 end-to-end RAG implementations — LangChain+ChromaDB, LangChain+Pinecone (PDF input), and LlamaIndex+FAISS+Groq. Multi-query retrieval, vector embeddings, and custom safety prompt templates for accurate document Q&A.

Multi-agent travel advisor using CrewAI where a Researcher agent and Planner agent collaborate to generate personalised itineraries with live weather, hotel, and attraction data via real-time web search.