PAL
Built as a personal project to assist with loneliness and to serve as a modular starter template for more advanced AI projects, PAL is a desktop voice assistant. It is designed to run both locally and online to support chats, questions, and deep conversational exchanges. Architected using Tauri, Rust, llama.cpp, and Groq APIs, the project runs as a background launcher. While it remains a perpetual work-in-progress, it already serves its core architectural purpose: providing the foundational STT/TTS template for other projects (like Numo).
About This Project
Creating a fast, highly private desktop assistant for speech-based companionship and coding support, while avoiding the privacy hazards of sending conversational contexts to corporate clouds.
Developed a background voice/chat client using Tauri and Rust. Integrated llama.cpp and Groq APIs to support local model execution and speech pipelines.
Created a lightweight, modular voice client that runs independently and serves as a boilerplate for subsequent AI integration (such as the Numo learning system).
AI & Desktop Developer
2024
Public
Personal
Technology Stack
Project Story
Creating a fast, highly private desktop assistant for speech-based companionship and coding support, while avoiding the privacy hazards of sending conversational contexts to corporate clouds.
Developed a background voice/chat client using Tauri and Rust. Integrated llama.cpp and Groq APIs to support local model execution and speech pipelines.
Created a lightweight, modular voice client that runs independently and serves as a boilerplate for subsequent AI integration (such as the Numo learning system).
Insights & Takeaways
Highlights
- Personal assistant built for companionship, supportive voice chatting, and coding queries.
- Integrates llama.cpp and Groq APIs for dual local and cloud inference.
- Serves as the core STT/TTS and model launcher boilerplate for subsequent projects.
Challenges
- Securing offline local model configurations using llama.cpp within a lightweight Tauri context.
- Building a low-latency voice model communication pipeline that behaves naturally during conversation.
Lessons Learned
- Proved that a custom, local first voice assistant successfully preserves user privacy and functions as an exceptional template for scaling other AI pipelines.
Related Work
Case Study
Auto Shutdown
Rust + React desktop app to automate PC shutdowns by timer, schedule or idle state.
Case Study
Azkari
Lightweight desktop app that displays Azkar reminders based on time of day, built with Tauri, SolidJS, TypeScript and Rust.
Case Study
EBEC
A full stack, highly creative web platform built for the ENSIA Business & Entrepreneurship Club, complete with an administration and user management system.
