Deepesh Agrawal

Architecting and developing AI-powered voice agents using LiveKit for real-time telephony and voice interactions. Built a comprehensive Next.js 15 dashboard with React 19 and TypeScript for managing agent lifecycles, real-time call monitoring, and SSO authentication. Developed serverless architectures on AWS Lambda with DynamoDB and SQS for scalable campaign management.

Conducted research in AI and machine learning at Samsung Prism, working on specific research topics including LLM engineering, advanced AI applications, and machine learning model optimization. Contributed to innovative research projects that advanced the state-of-the-art in AI systems.

B.Tech in Computer Science with CGPA: 8.24/10. Specialized in software development, AI/ML, and competitive programming. Won multiple hackathons including Hackfest IIT Dhanbad and Google Gen-AI Hackathon. Published 2 independent research papers in Springer conferences.

An intelligent AI agent that automatically generates comprehensive technical documentation by analyzing code structure, PR history, and codebase context. Leverages RAG (Retrieval-Augmented Generation) to understand code semantics and generate accurate, context-aware documentation. Built with TypeScript, OpenAI GPT-4, and advanced NLP techniques.

A Gen-AI powered cross-platform mobile application for creating comprehensive NPTEL courses with automated content generation. Built with React Native and Sarvam AI, featuring regional language support for inclusive education. Automatically structures curriculum, generates course materials, creates interactive learning modules, and produces assessment questions.

An AI-powered video learning platform that won Hackfest IIT Dhanbad. A platform that allows users to ask natural language questions about video content and receive accurate, timestamped results. Features both frame-based (I-SeekR) and audio-based (A-SeekR) search capabilities using machine learning algorithms. Built with Flask, React.js, and JavaScript for intelligent video content retrieval.

A secure, decentralized document signing platform leveraging blockchain technology for tamper-proof document authentication. Built with Solidity smart contracts on Ethereum, ensuring immutability and cryptographic verification of signatures.

Led development as Project Head of the official website for IIIT NR's annual technical festival, serving thousands of participants. Built a comprehensive event management system with real-time registration, payment integration, and admin dashboard. Implemented with Next.js, TypeScript, and Tailwind CSS.

Contributed to the Student Activity Center website for IIIT NR, serving as the central hub for student activities, event announcements, and information. Built with React.js, Node.js, and MongoDB.

Published research paper in Springer conference proceedings on Blockchain Technology. This paper introduces innovative authentication measures for document security by exploring the integration of multiple signature authentication and time-constrained signature features. The research addresses vulnerabilities in conventional digital signature methods by leveraging blockchain's immutability and cryptographic properties.

Published research paper in Springer conference proceedings on Computer Vision and Deep Learning. This paper presents a system that utilizes computer vision techniques to identify suspicious behavior in high-risk areas. The technology employs cameras to detect individuals engaged in suspicious activity or situated in hazardous locations, analyzing video using deep learning and computer vision techniques.