This course teaches how to build "Ava," a WhatsApp agent capable of realistic conversations, voice interaction, and image processing, inspired by the movie Ex Machina. It targets ML Engineers and Software Engineers looking to create complex, production-ready AI applications, offering a deep dive into LangGraph workflows and end-to-end agent development.

How It Works

Ava leverages LangGraph for complex workflow orchestration, integrating multiple AI models. It utilizes Qdrant for long-term memory, Whisper for speech-to-text, ElevenLabs for text-to-speech, and diffusion models (like FLUX) for image generation, with vision models (like Llama 3.2 Vision) for image understanding. The agent connects to the WhatsApp API and can be deployed on Google Cloud Run.

Quick Start & Requirements

• Installation and setup details are provided in a separate GETTING STARTED.md document.
• Prerequisites include setting up virtual environments, installing dependencies, and creating a .env file.
• Free tiers of services like Groq, ElevenLabs, Qdrant Cloud, and Together AI are sufficient for initial use. Google Cloud Run offers free credits for deployment.

Highlighted Details

• Build a fully functional WhatsApp agent with voice and image capabilities.
• Learn to implement LangGraph workflows for complex agentic applications.
• Set up long-term memory using Qdrant and integrate STT/TTS with Whisper and ElevenLabs.
• Implement image generation with FLUX models and image processing with VLM models.

Maintenance & Community

The project is associated with The Neural Maze, founded by Miguel Otero Pedrido, with contributions from Jesús Copado. Both are Senior ML/AI Engineers. Links to their LinkedIn and YouTube profiles are provided.

Licensing & Compatibility

Licensed under the MIT License, permitting commercial use and integration with closed-source projects.

Limitations & Caveats

The course assumes a certain level of technical proficiency for setup and development. Specific details on potential limitations or alpha status are not explicitly mentioned in the README.