muhammederem Chief Orchestration Framework
Chief is a powerful multi-agent orchestration framework that automates complex workflows across ML/AI, backend development, and DevOps.
When to Use
Use Chief when you need to:
- Train or fine-tune ML models (PyTorch, Hugging Face, LangChain)
- Build backend APIs (FastAPI, GraphQL, REST)
- Deploy to cloud (AWS Lambda, SageMaker, ECS)
- Create CI/CD pipelines (GitHub Actions, GitLab CI)
- Generate infrastructure (Terraform, Docker, Kubernetes)
- Build RAG applications with vector databases
Installation
git clone https://github.com/muhammederem/chief.git
cd chief
pip install -e .
Quick Commands
ML/AI Workflows
# Train a PyTorch model
chief exec /train model_type=cnn dataset=/data/images epochs=50
# Fine-tune Hugging Face model
chief exec /finetune model=bert-base-uncased dataset=custom_data output_dir=/models/bert
# Build RAG pipeline
chief exec /rag llm=gpt-4 vector_db=pinecone data=/documents
Backend Development
# Generate FastAPI project
chief exec /api project_name=my_api framework=fastapi features=auth,database
# Add Docker configuration
chief exec /dockerize app_type=fastapi port=8000 include_compose=true
# Integrate database
chief exec /database type=postgresql orm=sqlalchemy
DevOps & Deployment
# Deploy to AWS Lambda
chief exec /deploy service=lambda app_name=processor runtime=python3.9
# Create CI/CD pipeline
chief exec /pipeline platform=github stages=test,build,deploy
# Generate Terraform config
chief exec /terraform provider=aws resources=ec2,s3,rds region=us-west-2
Project Templates
Generate complete project scaffolds:
# List available templates
chief template list
# Generate PyTorch project
chief template generate pytorch-starter my_project
# Generate FastAPI microservice
chief template generate fastapi-microservice my_api
# Generate MLOps pipeline
chief template generate mlops-pipeline my_pipeline
Key Features
- Multi-Agent System: ML Engineer, Backend Architect, DevOps Engineer
- Intelligent Routing: Automatically routes tasks to the right agent
- Domain Knowledge: Auto-loads skills for PyTorch, Hugging Face, FastAPI, AWS, etc.
- Project Templates: Ready-to-use scaffolds for common architectures
- Load Balancing: Work-stealing, round-robin strategies for parallel execution
Documentation
See docs/ for complete documentation.