Prasannaa Kumar

Junior AI Engineer and Computer Science graduate with hands-on internship experience building and shipping production-grade Agentic AI systems, LLM-powered automation pipelines, and Retrieval-Augmented Generation (RAG) solutions for real business clients. Proficient in Python, LangChain, CrewAI, AutoGen, LangGraph, OpenAI GPT-4, OpenAI Assistants API with Function Calling, Pinecone, ChromaDB, LlamaIndex, FastAPI, REST APIs, and n8n workflow automation. Successfully delivered two live end-to-end AI deployments — an AI restaurant chatbot (RAG + WhatsApp + UPI payment) and an agentic lead qualification pipeline with Gmail and Telegram integration.

Strong in prompt engineering, multi-agent orchestration, vector database integration, and no-code/low-code AI automation using Make.com and Zapier. Passionate about building intelligent AI systems that solve real business problems.

AI Engineer Intern — Agentic AI & LLM Automation at Omnilight AI Solutions (2025-07 – 2025-12)

6-month internship building production AI automation systems using LLMs, RAG pipelines, and multi-agent orchestration for real business clients.

• Designed and deployed end-to-end agentic AI workflows using Python, n8n, GPT-4, and LangChain — automating lead qualification, scoring, and intelligent routing via multi-step LLM reasoning pipelines for real business clients
• Built and integrated RAG pipelines using Pinecone and ChromaDB to ground LLM responses in structured business data — reducing hallucinations by 40% and improving response accuracy across all deployed systems
• Architected multi-agent orchestration systems using LangChain ReAct decision loops — enabling autonomous tool selection, dynamic API triggering, and conditional action execution across integrated platforms
• Developed OpenAI Assistants API integrations with Function Calling and JSON schema enforcement — ensuring consistent, parseable structured outputs for all downstream n8n and FastAPI automation nodes
• Built REST API integrations and FastAPI-based endpoints to connect AI workflows with third-party services — enabling real-time data exchange between LLM systems and external platforms
• Implemented Prompt Engineering strategies (zero-shot, few-shot, iterative optimization) — improving LLM classification accuracy and reducing lead false positives by 35% across automated pipelines
• Integrated 6+ third-party APIs (Gmail, Google Sheets, Telegram, WhatsApp Business API, Google Calendar, UPI Payment) — eliminating manual data entry and reducing client processing time by 70%
• Collaborated with stakeholders to identify automation use cases, define AI system architecture, and deliver complete solutions from design through production deployment and monitoring

Bachelor of Engineering in Computer Science & Engineering – Jeppiaar Institute of Technology (2021 – 2025)