AI Prompt Engineer and Fullstack Developer creating intelligent, autonomous systems. From engineering complex AI agents to building scalable fullstack applications, I turn cutting-edge ideas into reality.
View My WorkLeverages the powerful Gemini API for nuanced, context-aware message generation. Features a clean, responsive UI with a dynamic theme switcher. Implements client-side rate limiting to prevent API abuse and built with modern TypeScript for a type-safe, maintainable codebase.
View ProjectDeveloped a RL model using BART to extract key data & summaries from PDFs, continuously improving via feedback. Implemented SpaCy for advanced NER and text extraction using fitz, achieving 99.22% accuracy. Optimized data extraction workflows using Reinforcement Learning and an efficient deep learning pipeline.
Secured 3rd place by developing a CNN-based Convolutional Autoencoder for underwater image transmission. Processed over 8,000 images, utilized t-SNE for dimensionality reduction, and achieved over 85% reconstruction accuracy.
View on KaggleAn AI system to extract cancer-related information from patient Electronic Health Records (EHR). It features a multi-stage pipeline for information retrieval using keyword and semantic search, and a quantized LLM (Qwen-7B) to extract structured medical data into JSON format.
View on GitHubAn LLM-based model that allows users to query YouTube videos. It extracts and transcribes video audio using OpenAI's Whisper, then leverages a language model to provide summaries and answer specific questions about the video's content.
A real-time chat application featuring a secure MERN stack architecture. Engineered with a Node.js/Express backend using JWT for authentication and Socket.IO for instant, bi-directional communication. The React frontend provides a dynamic user experience, while MongoDB stores message history and user data efficiently.
View on GitHubA MERN app with secure authentication, admin/customer panels, and Stripe integration. Developed RESTful APIs using JWT for auth and bcrypt for hashing. Designed MongoDB schema with Mongoose and built a React frontend with Redux Toolkit and protected routes. Integrated Stripe Webhooks for payment validation.
View on GitHubAn AI workbench to build and perfect document extraction prompts. Auto-generates robust XML/text prompts from simple descriptions, intelligently refines failing prompts using granular feedback, and features a multi-document sandbox for rapid testing. Leverages Gemini's JSON mode with strict schemas for reliable, structured data output.
View ProjectRefines AI-generated text to sound exceptionally human, preserving meaning and formatting. Employs a multi-stage AI pipeline for deep text humanization, using adversarial techniques to eliminate robotic sentence structures. Features an integrated AI content detector to score output quality and a diff viewer to show changes.
View ProjectAn AI-powered library to create, manage, refine, and version LLM prompts. Engineered a non-blocking Firestore architecture for optimistic UI updates and real-time collaboration. Developed a type-safe AI flow system with Genkit and implemented a dynamic variable system for interactive prompt testing.
View ProjectProvides instant, multi-layered AI summaries for any movie/TV series. Built a responsive UI with Next.js and Tailwind. Engineered AI flows with Genkit & Gemini to generate 3 distinct levels of plot summary and integrated with the TMDB API for comprehensive title info. Enhanced AI accuracy with a real-time web search tool.
View ProjectA secure, multi-tenant secrets management frontend powered by Google Cloud Secret Manager & Firebase. Engineered with namespaced secrets in GCSM and Firestore, achieving robust multi-tenancy. Decoupled metadata enables rapid UI operations. Built on Next.js with JWT-secured API routes and atomic folder deletion for data integrity.
View ProjectTransforms a raw app concept into a developer-ready technical brief using AI. Built a multi-step AI workflow that guides users to a full engineering plan. Implemented a dynamic AI questionnaire to refine user requirements and a system that intelligently pre-configures advanced settings based on user input.
View ProjectTubeQuery is a Python-based tool that leverages Large Language Models (LLMs) and advanced Natural Language Processing (NLP) techniques to enable users to extract information from YouTube videos. By simply providing a YouTube video URL, users can obtain a transcript, a concise summary, and answers to specific questions about the video's content. This project demonstrates the practical application of speech-to-text conversion, text summarization, and question-answering models in a real-world scenario.
Read Paper View on Kaggle
HackerRank
Skills: Python (Programming Language) · Software Development
Udemy
Skills: Deep Learning · Data Science · Machine Learning
Skills: Data Ethics · Spreadsheets · SQL · Data Analysis
Udemy
Skills: REST APIs · React.js · Node.js · Express.js · Full-Stack Development · MongoDB
Powered by Gemini