Platform Architecture Overview | Kamiwaza Docs
Documentation for Kamiwaza 0.7.0
This document provides a high-level overview of its key architectural components and the technologies that power them.
System Architecture Diagram
The diagram below illustrates the layered architecture of the Kamiwaza platform, from the underlying infrastructure to the user-facing applications. Each layer provides a distinct set of capabilities, creating a robust and maintainable system.
Consumes
Built On
Runs On
| Layer | Technologies |
|---|---|
| Orchestration & Compute Foundation | Ray Serve, Docker Swarm, Hardware / Cloud |
| Core Infrastructure Services | Traefik Router, CockroachDB, etcd Cluster |
| Kamiwaza Services Layer | Microservices, Authentication, Model Repository, Model Serving, Vector Databases, Data Catalog, FastAPI Gateway |
| Application Layer | Kamiwaza Frontend, Client SDK, App Garden |
Architecture Layers
🌐 Application Layer
This is where users and developers interact with Kamiwaza.
- Kamiwaza Frontend: The primary web-based user interface for managing the platform.
- Client SDK: A Python SDK for programmatically interacting with Kamiwaza's APIs.
- App Garden: A platform for discovering and deploying pre-packaged AI applications and services.
🎯 Kamiwaza Services Layer
The core business logic of the platform, exposed via a central API gateway.
- FastAPI Gateway: A high-performance API gateway that coordinates all requests.
- Microservices: A suite of specialized services handling concerns like Authentication, Model Management, Vector DB abstraction (Milvus, Qdrant), and more.
🛠️ Core Infrastructure Services
The essential backend services that support the entire platform.
- Traefik Router: A powerful reverse proxy and load balancer that manages all inbound network traffic and provides SSL termination.
- CockroachDB: The primary database—a distributed, resilient SQL database for all application data.
- etcd: A distributed key-value store used for service discovery and critical configuration management across the cluster.
🔧 Orchestration & Compute Foundation
The base layer that provides the compute resources and orchestration to run everything else.
- Ray Serve: A scalable model-serving framework used for deploying and managing AI models.
- Docker Swarm: The container orchestration engine that manages the lifecycle of all containerized services.
- Hardware / Cloud: The physical servers or cloud instances (e.g., AWS, GCP, Azure) that provide the underlying CPU and GPU resources.
Technology Stack
| Category | Technologies |
|---|---|
| Backend | Python 3.10, FastAPI, Ray, SQLAlchemy, Pydantic |
| Frontend | React 18, Material-UI, Tailwind CSS, Axios |
| Databases | CockroachDB, Milvus, Qdrant, etcd |
| Infrastructure | Docker Swarm, Traefik, DataHub, JupyterHub |