Priya Kumari

AI Engineer specializing in Agentic AI workflows, intelligent document processing (IDP), LLM-Powered and cloud-native AI systems. Experienced in orchestrating multi-agent pipelines using LangGraph, LangChain, Azure, and advanced prompting techniques, transforming unstructured data into actionable business insights.

Full Stack LLM Development Analyst - Accenture

(2022-01)

• Evaluated AWS Textract + Claude Sonnet vs Azure Document Intelligence + GPT-4o for invoice/receipt processing, contributing to Azure tech stack selection.
• Built a LangGraph-powered agentic workflow orchestrating multiple autonomous agents (OCR, validation, GPT reshaping, translation) for semi-structured and unstructured complex data from documents, reducing processing time ~30% via parallel execution.
• Developed a PII masking Pipeline using Azure Text Analytics (PII + NER APIs), achieving >90% accuracy and redacting sensitive data exposure to LLMs.
• Crafted GPT-4 prompts using zero-shot, few-shot, and chain-of-thought (CoT) prompting strategies to enhance data extraction accuracy and contextual consistency.
• Deployed Azure Functions and REST APIs (Python / Node.js) with Cosmos DB integration and CI/CD pipelines ensuring 99%+ reliability.
• Developed an enterprise travel assistant using Azure Functions (Python & Node.js) integrating travel booking APIs.
• Implemented a RAG pipeline using Azure OpenAI and Azure AI Search for context-aware travel assistance.
• Secured REST APIs using Microsoft Entra ID, OAuth2, JWT, and Azure Key Vault.
• Designed asynchronous workflows using Azure Queue Storage and Azure Table Storage.
• Automated infrastructure and deployments using Terraform and Azure DevOps while supporting API integration and security remediation.

Masters Of Computer Application - MCA - Kalinga Institute of Industrial Technology (2020-01 - 2022-12)