Azure Service Intelli Debugger - AI Engine

This is the production-ready neural engine for autonomous Logic App error remediation. It uses a 6.8M parameter Transformer model optimized for Apple Silicon (MPS).

🚀 Quick Start: Training the Model

To train the model on your local data:

1. Navigate to Core:

   cd ai_engine/core
   

2. Run Training:

   python3 train.py --data ../data/comprehensive_training_data.json --num-epochs 2 --batch-size 8 --pre-tokenize
   

🧪 Testing the Model

Via Command Line

You can pipe a test JSON into the inference engine:

python3 inference.py < test_samples.json

Via VS Code AI Playground

1. Open the AI Model Performance Playground in VS Code.
2. Select "Isolated AI (PyTorch)" as the provider.
3. Copy a prompt from ai_engine/core/test_samples.json.
4. Paste it into the text box and click Run Performance Test.

📂 Project Structure

• ai_engine/core/: The neural engine (Model, Training, Inference).
• ai_engine/data/: High-quality training datasets.
• src/: VS Code extension source code.

🛠️ Tech Stack

• Engine: PyTorch (Metal Performance Shaders for Mac).
• Architecture: decoder-only Transformer with RoPE and SwiGLU.
• Integration: TypeScript-based VS Code Extension.