runnable 0.50.0

Add your description here

Runnable

Transform any Python function into a portable, trackable pipeline in seconds.

---

🚀 30-Second Transformation

Your existing function (unchanged!):

def analyze_sales():
    total_revenue = 50000
    best_product = "widgets"
    return total_revenue, best_product

Make it runnable everywhere:

from runnable import PythonJob

def main():
    PythonJob(function=analyze_sales).execute()

if __name__ == "__main__":
    main()

🎉 Success! Your function now runs the same on laptop, containers, and Kubernetes with automatic tracking and reproducibility.

🔗 Chain Functions Without Glue Code

def load_customer_data():
    return {"count": 1500, "segments": ["premium", "standard"]}

def analyze_segments(customer_data):  # Name matches = automatic connection
    return {"premium_pct": 30, "growth_potential": "high"}

# What Runnable needs (same logic, no glue):
from runnable import Pipeline, PythonTask

def main():
    Pipeline(steps=[
        PythonTask(function=load_customer_data, returns=["customer_data"]),
        PythonTask(function=analyze_segments, returns=["analysis"])
    ]).execute()

if __name__ == "__main__":
    main()

Same pipeline runs unchanged on laptop, containers, and Kubernetes.

⚡ Installation

pip install runnable

For development:

uv sync --all-extras --dev

Run examples:

uv run examples/01-tasks/python_tasks.py

📊 Why Choose Runnable?

• 🎯 Easy to adopt: Your code remains as-is, no decorators or imposed structure
• 🏗️ Bring your infrastructure: Works with your platforms, not a replacement
• 📝 Reproducibility: Automatic tracking without additional code
• 🔁 Retry failures: Debug anywhere, retry from failure points
• 🧪 Testing: Mock/patch pipeline steps, test functions normally
• 💔 Move on: Easy removal - just delete runnable files, your code stays

🎯 Core Strengths & Extensibility

Runnable excels at data pipeline orchestration:

• Data processing pipelines
• ML model training workflows
• ETL operations
• Batch job orchestration
• Scientific computing reproducibility

🧪 Extensible Architecture & Experimental Features:

Runnable's plugin-based architecture demonstrates its extensibility through experimental features like:

• Async streaming capabilities - Proof of concept for real-time processing (local execution)
• Loop workflows - Dynamic iteration patterns for complex data processing
• Custom executors - Extensible to any infrastructure via plugins

For production agentic frameworks, consider specialized tools like Pydantic AI, LangChain, or CrewAI, which are purpose-built for complex LLM applications.

📖 Documentation

Complete Documentation →

🔀 Pipeline Types

Linear Pipelines

Simple sequential execution of Python functions, notebooks, or shell scripts.

Parallel Branches

Execute multiple branches simultaneously for improved performance.

Map Patterns

Execute pipelines over iterable parameters for batch processing.

Arbitrary Nesting

Combine parallel, map, and sequential patterns as needed.

🆚 Why Choose Runnable?

	Runnable	Kedro	Metaflow	Airflow
Zero Code Changes	✅ Wrap existing functions	❌ Restructure to nodes	❌ Convert to FlowSpec	❌ Rewrite as DAG tasks
Environment Portability	Same code: laptop→container→K8s→Argo	Deployment-specific configs	AWS-focused with --with flags	Platform-specific operators
Mixed Task Types	Python + Notebooks + Shell	Python nodes only	Python steps only	Requires separate operators
Plugin Extensibility	Auto-discovery via entry points	kedro-* packages	Limited extensions	Complex plugin development
Parameter Passing	Automatic by name matching	Manual catalog definitions	Flow state management	Manual XCom operations

🚀 Time to Value

Runnable:

pip install runnable
# Wrap your existing function - done!

Others:

• Learn framework conventions
• Restructure existing code
• Configure project structure
• Set up deployment configs

---

Ready to get started? Check out our 30-second demo for immediate results!