Documentation for Kamiwaza 0.9.3

This documentation describes the Kamiwaza platform, composed of several key components and conceptual systems that work to provide a comprehensive AI orchestration solution.

Model & Data Handling

Vector Databases

• What it is: A specialized database for storing and retrieving high-dimensional vector data, such as embeddings generated from text or images. Kamiwaza integrates with industry-standard vector databases like Milvus and Qdrant.
• Why it matters: Vector databases are the engine behind powerful similarity searches, essential for Retrieval-Augmented Generation (RAG), recommendation engines, and other advanced AI applications. Kamiwaza's abstraction layer lets you choose the right database for your needs without changing your application code.

Embeddings Management

• What it is: The process of generating, storing, and managing the vector embeddings for your data. Kamiwaza provides built-in services to automate the creation of embeddings using various open-source or custom models.
• Why it matters: Consistent and efficient embedding management is fundamental to the performance of any vector search-based application. By handling this automatically, Kamiwaza reduces a major source of complexity in building RAG pipelines.

Data Catalog

• What it is: A centralized inventory of all your data assets. Kamiwaza integrates with Acryl DataHub to provide a single place to discover, understand, and govern your data.
• Why it matters: As AI systems grow, so does the data they consume. A data catalog provides crucial lineage tracking and discoverability, which is vital for enterprise governance, security, and scalability.

Orchestration & Serving

Orchestration Engine

• What it is: The "brain" of the platform that manages the flow of requests and coordinates tasks between different services. Kamiwaza uses frameworks like Ray Serve to handle this complex, distributed workload.
• Why it matters: The orchestrator ensures that AI requests are processed efficiently, scaled according to demand, and routed to the correct models and services. This is the key to building resilient, production-grade AI applications.

Model Serving

• What it is: The process of taking a trained AI model and making it available for real-time inference via an API. Kamiwaza supports multiple high-performance serving engines like vLLM, llama.cpp, and MLX.
• Why it matters: Different models have different hardware needs. Kamiwaza's multi-engine support ensures you can run a diverse range of models and optimize for both performance and cost.

API Gateway

• What it is: A single, unified entry point for all API requests to the platform. Kamiwaza uses FastAPI to create this gateway, which then routes requests to the appropriate internal microservice.
• Why it matters: A gateway simplifies development by providing a consistent interface for all platform services. It's also the ideal place to enforce cross-cutting concerns like authentication, rate limiting, and logging.

Security & Operations

Identity & Access Management

• What it is: The system that handles user authentication and authorization.
• Why it matters: Robust security is non-negotiable in an enterprise setting. Kamiwaza's IAM services ensure that only authorized users and applications can access sensitive data and models.

Caching

• What it is: A high-speed storage layer (e.g., Redis, Valkey) that keeps frequently accessed data readily available, reducing the need to re-compute or re-fetch it from slower databases.
• Why it matters: Caching dramatically improves the performance and responsiveness of AI applications, especially those with high request volumes, leading to a better user experience and lower operational costs.