kailash-kaizen 0.2.0

Advanced AI agent framework built on Kailash SDK

Kaizen AI Framework

Signature-based programming framework for enterprise AI workflows

Kaizen is an AI framework built on the Kailash SDK that provides signature-based programming, automatic optimization, and enterprise-grade capabilities.

Quick Navigation

📁 Project Organization

• /docs/ - Technical documentation, architecture, and implementation guides
• /examples/ - Working examples with complete implementations
• /adr/ - Architecture Decision Records and technical rationale
• /tracking/ - Implementation progress and gap analysis
• /scripts/ - Development tools and validation utilities
• /src/ - Source code implementation
• /tests/ - Comprehensive test suite

🎯 Key Documents

• CLAUDE.md - Claude Code navigation and agent instructions
• docs/README.md - Documentation hub with organized guides
• Multi-Modal API Reference - Vision, audio, multi-modal APIs (NEW ✅)
• Integration Testing Guide - Real model validation best practices (NEW ✅)
• examples/README.md - Example categories and usage patterns
• tracking/ - Current implementation status and roadmap

Features

• Signature Programming: Declarative AI workflow definition
• Enterprise Ready: Built on proven Kailash infrastructure
• Auto-Optimization: ML-based prompt and workflow improvement
• Multi-Modal AI: Vision (Ollama/OpenAI), audio (Whisper), unified orchestration (Phases 0-4 ✅)
• Production Scale: Performance and reliability for enterprise use
• Cost Tracking: Monitor and optimize AI API usage with budget management

Quick Start

from kaizen import Kaizen
from kailash.runtime.local import LocalRuntime

# Initialize framework
kaizen = Kaizen()

# Create AI agent with signature-based programming
agent = kaizen.create_agent("text_processor", {
    "model": "gpt-4",
    "temperature": 0.7
})

# Execute with Core SDK runtime
runtime = LocalRuntime()
results, run_id = runtime.execute(agent.workflow.build())

Installation

pip install kailash-kaizen

For development:

pip install kailash-kaizen[dev]

Architecture

Kaizen is built ON the Kailash SDK, providing:

• Signature-based programming layer
• Enhanced AI nodes with optimization
• Enterprise memory and security features
• Seamless integration with DataFlow and Nexus

Developer Experience

Documentation Structure

• Getting Started: See examples/README.md for quickstart patterns
• Multi-Modal AI: See docs/reference/multi-modal-api-reference.md for vision, audio, and multi-modal APIs
• Integration Testing: See docs/development/integration-testing-guide.md for real model validation
• Technical Details: See docs/README.md for comprehensive guides
• Implementation Status: See tracking/ for current progress and gaps
• Architecture Decisions: See adr/ for design rationale and requirements

Development Workflow

1. Explore Examples: Start with examples/1-single-agent/ for basic patterns
2. Check Implementation: Review tracking/KAIZEN_GAPS_ANALYSIS.md for current status
3. Read Architecture: Study docs/KAIZEN_INTEGRATION_STRATEGY.md for framework integration
4. Run Tests: Use scripts/test_examples_implementation.py for validation

Contributing

See CONTRIBUTING.md for guidelines.

Links

• Repository: GitHub
• Documentation: Full Documentation
• PyPI: kailash-kaizen
• Issues: Bug Reports & Feature Requests

License

Apache-2.0 WITH Additional-Terms - See LICENSE for details.