Student

github

medium

Om Sudhamsh  • Peerlist

Secure Link Generator

<h3>An AI-powered prompt engineering tool</h3>This utility helps users create and manage optimized prompts for various AI models and applications. It leverages Google Gemini AI to generate prompts with different tones and styles, tailored for specific tasks such as Data Analytics, CSV Processing, Image Creation, and Task Planning.<ul><li>AI-Powered Generation: Utilizes Google Gemini AI for intelligent prompt optimization.</li><li>Multiple Task Types: Supports Data Analytics, CSV Processing, Image Creation, and Task Planning.</li><li>Tone Selection: Offers a variety of tones including Professional, Casual, Persuasive, Friendly, Formal, Humorous, and Concise.</li><li>Optimization Scoring: Provides a real-time score (0-100) indicating prompt quality based on clarity, length, and structure.</li><li>Smart Suggestions: Offers AI-generated recommendations for prompt improvement.</li><li>File Upload Support: Allows users to upload context files (e.g., CSV) to enhance prompt generation.</li><li>Export Options: Enables downloading generated prompts in TXT or Markdown formats.</li><li>Chat History: Keeps track of previous prompt generations for review.</li><li>Responsive Design: Features a mobile-friendly interface built with Tailwind CSS.</li></ul>The project includes a backend developed with FastAPI and a frontend built using React, deployed via GitHub Pages for a live demo and Vercel for full-stack deployment.

Prompt Generator

This project is a Python-based system designed to detect duplicate questions on Stack Overflow. It utilizes advanced machine learning techniques, specifically Convolutional Neural Networks (CNN) combined with Natural Language Processing (NLP), to accurately identify semantically similar queries. The system features an interactive Flask-based web interface that allows users to easily submit questions for analysis, process multiple questions in batches, and train the model directly through the UI with progress tracking. It provides real-time predictions and displays detailed model performance statistics. For user convenience, sample questions from Stack Overflow are included, and the model can be trained in the background using threading to avoid UI blocking. Prerequisites include Python 3.7+, pip, and Git.<ul><li>Smart Duplicate Detection: Uses CNN-based NLP models to identify duplicate questions</li><li>Web UI: Interactive Flask-based web interface for easy question submission and analysis</li><li>Batch Processing: Submit and analyze multiple questions at once</li><li>Model Training: Train the model directly from the web interface with progress tracking</li><li>Real-time Predictions: Get instant results for your queries</li><li>Model Statistics: View detailed model performance and training metrics</li><li>Sample Questions: Try the model with pre-loaded sample questions from StackOverflow</li><li>Background Training: Train the model without blocking the UI using threading</li></ul>

Duplicate Question Detection using DL

<h3>Project Description</h3>This project implements a Convolutional Neural Network (CNN) to recognize CAPTCHA images. It's designed as a learning project for the RTP-B14 team. The goal of this project is to build a CNN model that can recognize characters in CAPTCHA images. The model is trained to predict individual characters in the CAPTCHA, handling both letters and digits.<h3>Dataset</h3>The project uses a custom dataset of CAPTCHA images. Each image contains 5-6 characters (both letters and digits) with some noise and distortion typical of CAPTCHAs.<h3>Requirements</h3><ul><li>Python 3.6+</li><li>TensorFlow 2.x</li><li>Keras</li><li>OpenCV</li><li>NumPy</li><li>Matplotlib</li></ul><h3>Installation</h3><ol><li>Clone the repository:</li></ol>git clone https://github.com/omsudhamsh/captcha-recognition-using-cnn-rtp-b14.git
cd captcha-recognition-using-cnn-rtp-b14

Captcha Recognition using CNN

<h3>AI Data Analyst</h3>Upload any CSV → Ask questions in plain English → Get SQL &amp; results instantlyThis web application allows users to upload CSV files and query them using natural language. The AI backend translates these queries into SQL statements, which are then executed against a SQLite database. The results are displayed in a clean, interactive table. Key features include dynamic schema detection from CSV files, one-click SQL copying, and a modern dark UI with smooth animations. For safety, only SELECT queries are permitted, and output is sanitized.<ul><li>CSV Upload — Upload any .csv file as your dataset</li><li>AI-Powered — Converts natural language to SQL using Groq's LLaMA 3.1</li><li>Instant Results — See query results in a clean, interactive table</li><li>Dynamic Schema — Auto-detects table structure from uploaded CSV</li><li>Copy SQL — One-click copy for generated queries</li><li>Keyboard Shortcut — Ctrl + Enter to submit</li><li>Premium Dark UI — Modern design with smooth animations</li><li>Safe Queries — Only SELECT statements allowed</li></ul>The architecture involves a React, Vite, and Tailwind CSS frontend deployed on Vercel, and a FastAPI (Python) backend deployed on Render. Data is stored in a SQLite database, and AI processing is handled by Groq's LLaMA 3.1 model.

Natural Language to SQL

<h3>A JavaScript-based chatbot implementation utilizing Retrieval-Augmented Generation (RAG) techniques for intelligent conversation handling.</h3>
This project is a retrieval-augmented generation chatbot that allows users to upload PDFs and ask questions using Google Gemini AI. The system processes uploaded documents by extracting text, converting it into vectors, storing these vectors in a database, and then retrieving relevant context to generate accurate and informative responses.
Key Features:
<ul>
<li>PDF Upload: Users can upload PDF documents to serve as the knowledge base for the chatbot.</li>
<li>Intelligent Q&amp;A: Ask questions about the content of the uploaded PDFs and receive contextually relevant answers.</li>
<li>RAG Implementation: Leverages Retrieval-Augmented Generation for enhanced response accuracy.</li>
<li>AI Integration: Utilizes Google Gemini AI for natural language understanding and response generation.</li>
<li>Vector Database: Stores document embeddings for efficient similarity search.</li>
</ul>
Tech Stack:
<ul>
<li>Backend: FastAPI, Python 3.12, Sentence Transformers, Google Gemini, Supabase, Uvicorn.</li>
<li>Frontend: React 19, Vite, Axios, Tailwind CSS.</li>
</ul>
Setup: The project includes detailed setup instructions for cloning, installing dependencies, configuring environment variables (including Supabase credentials and Gemini API key), and running both the backend and frontend applications.
Usage: Simply upload a PDF, type your question, and the chatbot will provide answers based on the document's content.