About Me
Computer Science and Engineering (AI & ML) graduate with hands-on experience developing AI-powered applications, intelligent agents, and Retrieval-Augmented Generation (RAG) systems using Python, FastAPI, LangChain, and LangGraph. Experienced in building LLM-driven workflows, prompt-engineering solutions, AI evaluation frameworks, and scalable backend services that solve real-world problems. During my internship at CYPWNG Software Technologies, I contributed to the Visa2Book platform by developing AI-powered recommendation workflows, FastAPI services, prompt-engineering solutions, and evaluation pipelines. I worked on AI feature development, data processing, model evaluation, and deployment workflows while collaborating in an agile development environment. My projects include an AI Memory Agent and a Multi-Agent Document Intelligence System, where I implemented AI agents, contextual memory, hybrid search, vector retrieval, and multi-step reasoning using modern Generative AI technologies. I enjoy exploring Large Language Models (LLMs), Agentic AI, Prompt Engineering, LangChain, LangGraph, Hugging Face, and RAG architectures. I am actively seeking opportunities as an AI Engineer, Generative AI Engineer, AI/ML Engineer, or LLM Engineer where I can contribute to building intelligent AI products, scalable AI systems, and next-generation agentic applications.
Skills & Technologies
Work Experience
AI Developer Intern
Developed and maintained responsive web application features using HTML, CSS, JavaScript, Python, and Django, contributing to the Visa2Book platform's frontend and backend functionality. Built data preprocessing and feature engineering pipelines using Pandas and NumPy to support AI-driven recommendation and decision-support workflows. Integrated AI-powered recommendation systems and implemented prompt-engineering workflows using LLM technologies to enhance user guidance and automation capabilities. Developed and optimized FastAPI-based inference services, containerized applications with Docker, and supported scalable deployment environments. Created automated evaluation, testing, and regression monitoring frameworks using Python and PyTest to track model performance, reliability, and continuous improvement opportunities. Assisted in developing and testing AI agents using LangChain and LangGraph workflows, enabling automated reasoning, task execution, and contextual response generation across user interactions. Worked with structured and unstructured knowledge bases to support RAG-based AI applications, evaluating model outputs and refining prompt-engineering strategies to improve response accuracy and relevance. Collaborated on modular backend architecture, CI/CD workflows, and Git-based development practices to ensure maintainable, scalable, and production-ready applications.
Projects
AI Memory Agent
Built an intelligent AI Agent using Python, FastAPI, LangChain, and transformer-based LLM workflows to automate contextual user interactions and multi-step reasoning tasks. Integrated Gemini APIs / AI models with backend logic to enable natural language understanding, response generation, and task execution through prompt engineering techniques. Implemented evaluation workflows to assess response quality, context retention, and reasoning consistency across agent interactions, supporting iterative model improvement and performance analysis. Integrated Gemini-based LLMs and prompt-engineering strategies to optimize conversational quality and task completion effectiveness. Coordinated with team members to track requirements, manage task priorities, resolve project-related queries, and ensure timely completion of deliverables within defined deadlines.
Multi-Agent Document Intelligence System
Architected a multi-agent RAG workflow using LangGraph to orchestrate document ingestion, context retrieval, response generation, and evaluation agents. Implemented Hybrid Search by combining FAISS (semantic) and BM25 (keyword) vector stores to significantly improve retrieval accuracy for complex user queries. Integrated Groq-powered LLMs (Llama 3.1) to generate low-latency, highly grounded responses complete with precise, page-level source citations and multi-turn conversational memory. Built knowledge-base ingestion and retrieval pipelines leveraging FAISS vector stores and hybrid search techniques to improve information retrieval performance.
Education
B.Tech
CSE(Artificial Intelligence and Machine Learning)