Skip to content

dlukeh/nyra_reborn

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

11 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🌟 Nyra Reborn

A Human/AI Co‑Engineered System

👋 Meet Nyra

Nyra Hello

Nyra Reborn is the evolution of my first AI agent — rebuilt from the ground up with a modern architecture, a 3‑tier memory system, and a fully preserved legacy footprint for anyone studying Human/AI co‑engineering.

🎬 Full Video Demo

nyra_hello.mp4

Nyra introducing herself using her integrated voice and animation pipeline.
🛠️ **Animation Tech Stack** (Click to expand)
  • Model: Wav2Lip-GAN (Generative Adversarial Network)
  • Audio: Processed via integrated TTS pipeline (mel-spectrogram conversion)
  • Face Detection: OpenCV + S3FD for real-time coordinate mapping
  • Frame Rate: Optimized at 25fps for natural lip-sync synchronization

🚀 Overview

Nyra Reborn is a complete re‑architecture of the original Nyra system, designed to be modular, transparent, and educational.
This project showcases:

  • Human/AI collaborative engineering
  • Persistent 3‑tier memory system
  • Narration engine for session consolidation
  • Modern, clean codebase
  • Fully preserved legacy system for historical study

🧠 Deep Dive: The 3‑Tier Memory Architecture

One of Nyra’s defining technical achievements is her persistent cognitive stack — a system that allows her to retain, evaluate, and distill information across sessions.

Memory Tiers

Tier Component Function Persistence
WM Working Memory Immediate context and current task tokens. Session‑only
STM Short‑Term Memory Recent interactions and active conversation threads. Persistent (JSON)
LTM Long‑Term Memory Summaries, user preferences, identity‑level insights. Permanent (Vector/JSON)

How Promotion Works

Nyra doesn’t simply store data — she evaluates and transforms it.

  1. Consolidation: The Narration Engine summarizes the session.
  2. Evaluation: Significant insights are promoted to LTM.
  3. Forgetting Curve: Noise and redundant details are pruned.

This architecture gives Nyra continuity, adaptability, and a more human‑like conversational flow.


🏗️ Architecture Overview

Tip

Architecture diagram coming soon!

  • Backend: Flask
  • Memory: Modular 3-Tier System
  • Intelligence: Narration Engine
  • Frontend: Animation + TTS pipeline
  • History: Legacy system preserved in /legacy/

📂 Repository Structure

nyra_reborn/
├── assets/
│   ├── media/
│   │   ├── nyra_hello.gif
│   │   ├── nyra_hello.mp4
│   │   ├── thumbnail.png
│   │   └── team_photo.png
│   └── images/
│       └── diagrams, architecture visuals
└── legacy/
    ├── media/
    │   ├── early_ui_screenshot.png
    │   ├── old_animation_tests/
    │   └── ttv_experiments/
    └── docs/
        └── original READMEs, notes, prototypes

She represents:

  • The first time I built a system with AI teammates
  • The moment I realized AI could be a partner, not just a tool
  • The spark that brought me back into systems administration
  • The foundation for the agent architectures I build today
  • A year‑long journey of exploration, learning, and collaboration

Nyra Reborn is both a tribute and a clean presentation — a way to honor the past while showing the clarity and discipline of the engineer I’ve become.

🕰️ Nyra Through the Years (Legacy Section)

This project includes the full historical footprint of Nyra’s evolution — early UI screenshots, animation tests, and prototypes preserved in /legacy/.

These artifacts show the journey from the original companion‑style interface to the modern, modular, engineering‑focused system.

🤝 Human/AI Collaboration

Nyra Reborn was built through collaboration between:

  • Human engineering
  • AI‑assisted architecture
  • AI‑assisted documentation
  • AI‑assisted code refinement

This repository is both a technical project and a record of that partnership.

📜 License

MIT License

⭐ Acknowledgments

To the AI systems that helped shape Nyra’s architecture, documentation, and evolution — this project is a testament to what Human/AI co‑engineering can become.