TalkToTube

AI-Powered YouTube Q&A

TalkToTube is a Streamlit-based AI application that allows users to ask questions about YouTube videos and get accurate answers. It processes videos by downloading audio, transcribing it with Whisper, summarizing with BART, and retrieving relevant information using FAISS and BM25. The app then generates answers with Gemini AI.

With a user-friendly interface, efficient caching, and advanced AI models, TalkToTube makes it easy to extract insights from video content without watching the entire video.

Chaticle

AI-Powered Webpage Assistant

CHATICLE is a Chrome extension that makes browsing more efficient by allowing users to chat with any webpage. When enabled, it analyzes the entire page, letting users ask questions in natural language to quickly extract relevant information. Instead of manually searching through long, dense content, users get concise answers in a simple and interactive UI.

With CHATICLE, finding specific details on a webpage becomes effortless, saving time and enhancing productivity.

LLM Applications

AI Agnets and Retrieval Augmented Generation

-> AI-powered text, blog, and article generation.
-> AI-driven insights with interactive charts and graphs.
-> Intelligent system for finding and qualifying leads.
-> AI-based meme creator for social media engagement.
-> Enhances prompts for better LLM performance.
-> AI assistant for trip planning and itinerary generation.

Retrieval-Augmented Generation (RAG)
-> Agentic RAG – Uses intelligent agents to retrieve and generate contextually relevant responses.
-> Local RAG – Offline RAG system for privacy-focused and efficient document retrieval.

Fine-Tuning AI Models

Computer Vision Models
-> YOLO – Real-time object detection.
-> EfficientNet – High-performance image classification.
-> ResNet – Deep residual networks for image recognition.
-> SAM – Advanced image segmentation.

Text Generation Models
-> GPT – OpenAI's powerful text generator.
-> BERT – Contextual word embeddings for NLP.
-> LLaMA – Meta's open-source language model.
-> DeepSeek – Next-generation language understanding and generation model.

Speech Recognition Models
-> Whisper – OpenAI's robust automatic speech recognition (ASR) model.
-> DeepSpeech – Mozilla’s open-source speech recognition system.