Free deep-learning monitoring PWA for pets, babies, and elderly care. Real-time computer vision (TensorFlow.js COCO-SSD + Transformers.js ViT) runs entirely on-device, fully offline. Privacy-first.

Web component for removing image backgrounds in the browser

Cat & Dog Classifier is a browser-based machine learning application that classifies uploaded images as either cat or dog using a convolutional neural network. The project demonstrates how deep learning models can be integrated into a modern frontend stack using React, TypeScript, and TensorFlow.js, enabling real-time inference directly in the brow

AI-powered image upscaling and enhancement in the browser, built with React, TypeScript, and TensorFlow.js (ESRGAN & MAXIM models).

Neural Networks can learn (almost!) anything.

Private, on-device AI meeting notetaker that runs entirely in your browser — live transcription, speaker ID, and note extraction with no audio leaving your machine (WebGPU + WASM).

Innovative AI-powered music player built with Next.js 14. Integrates TensorFlow.js for real-time, browser-based hand gesture controls and the Spotify Web API for seamless audio streaming, backed by a robust NextAuth and MongoDB architecture.

A neural network learns to play a 2D platformer in real time — training, inference & visualization all in the browser via TensorFlow.js on WebGPU

An AI-powered, vibe-aware location discovery app combining ML, LLMs, and Maps.

Enterprise-grade NLU bot trainer with 5-classifier stacking ensemble (171K params) — runs entirely in the browser

Browser-based audio similarity search — no server, no uploads, runs entirely in your browser.

In-browser academic paper search engine with semantic search, knowledge graph visualization, and multi-source support (arXiv, OpenAlex, Europe PMC).

Interactive machine learning demos with TensorFlow.js featuring linear regression and binary classification with Materialize CSS framework

Client-side facial geometry and skin analysis using MediaPipe Face Mesh. 11 subscores from 468 3D landmarks, entirely in the browser. TypeScript, zero runtime dependencies.

In-browser neural network playground — train CNNs and dense nets on MNIST, Fashion-MNIST, and CIFAR-10 in real time

AI-powered fence detection & recoloring tool using UNet++ deep learning. Upload fence photos, apply realistic stain colors instantly.

An interactive, browser-based Real-Time Image Classifier using TensorFlow.js. Train a custom neural network entirely client-side using your webcam, complete with live predictions, model exporting.

GCS-SSC extension for narrative quality scoring

Benchmarking fine-tuned transformer models for multilingual sentiment analysis (EN/AR/RU) running entirely in the browser.