40 lines
1.3 KiB
Plaintext
40 lines
1.3 KiB
Plaintext
---
|
|
title: Jan Server
|
|
description: Self-hosted AI infrastructure running the Jan platform on Kubernetes.
|
|
keywords:
|
|
[
|
|
Jan Server,
|
|
self-hosted AI,
|
|
Kubernetes deployment,
|
|
Docker containers,
|
|
AI inference,
|
|
local LLM server,
|
|
VLLM,
|
|
Go API gateway,
|
|
Jan-v1 model
|
|
]
|
|
---
|
|
|
|
## Self-Hosted Jan Platform
|
|
|
|
Jan Server deploys the Jan AI platform on your own infrastructure using Kubernetes. It provides a complete AI inference stack with API gateway, model serving, and data persistence.
|
|
|
|
Jan Server is in early development. APIs and deployment methods may change.
|
|
|
|
## Architecture Overview
|
|
|
|
Jan Server consists of two main components:
|
|
|
|
- **API Gateway**: Go application handling authentication, web requests, and external integrations
|
|
- **Inference Model**: VLLM server running the Jan-v1-4B model for AI inference
|
|
- **PostgreSQL**: Database for user data, conversations, and system state
|
|
|
|
## Key Features
|
|
|
|
- **Kubernetes Native**: Deploys via Helm charts with minikube support
|
|
- **Jan-v1 Model**: 4B parameter model optimized for reasoning and tool use
|
|
- **OpenAI Compatible API**: Standard endpoints for integration
|
|
- **Authentication**: JWT tokens and OAuth2 Google integration
|
|
- **External Integrations**: Serper API for web search capabilities
|
|
- **Development Ready**: Local development environment with hot reload
|