JavaScript/TypeScript implementation of LLMLingua-2 (Experimental)

A Browser-Based AI Audio Transcription Playground Powered by Whisper.

A completely private, offline-capable AI chat interface where the language model runs entirely in your browser via WebGPU. Nothing leaves your device no servers, no API keys, no data transmission.

A floating developer notepad for Windows — built in Electron + React + TypeScript with offline on-device AI inference, sub-millisecond SQLite FTS5 search, and a multi-window IPC architecture. Zero network requests at runtime.

Small experiments for presentation

Production-grade speech data infrastructure platform for multilingual ASR/TTS training, featuring distributed processing, speaker diarization, emotion/style annotation, active learning, dataset optimization, observability and AWS-native deployment.

JustServiceHub is a full-stack web application that enables users to discover, book, and review local service providers. The platform supports multiple user roles (customers, providers, administrators) and includes AI-powered service recommendations, real-time location tracking, and secure payment processing.

Offline-first AI assistant (browser LLM + RAG) for newcomers & disconnected regions. No API keys, works in airplanemode.

Chat with a 1.2B Japanese LLM entirely in your browser — locally, privately, no backend needed.

Mainly just to be compatible with Obsidian plugins by brute force

🎥 | Aplicação web para resumir vídeos shorts do Youtube em formato de texto

An open-source On-Device AI powered study assistant that summarizes notes, generates flashcards, extracts keywords, and creates quizzes locally in the browser for complete privacy.

A privacy-first resume builder PWA running local LLMs (LFM-350M) via WebGPU & Transformers.js. Features an agentic chatbot that edits layouts & text in real time via tool calling, plus an on-device AI PDF parser.