Home â€ē Documentation â€ē Installation

Install NexusAI Engine

Get the NexusAI inference runtime and CLI up and running on your local machine or production infrastructure in under 5 minutes.

📋 Prerequisites

đŸ–Ĩī¸
OS: Linux (Ubuntu 20.04+), macOS 12+, or Windows 10/11 (WSL2)
⚙ī¸
Node.js 18+ or Python 3.10+ (depending on SDK)
đŸŗ
Docker 24+ & Docker Compose (for containerized deployment)
🔑
Valid NexusAI API Key from Developer Console
💾
Minimum 8GB RAM, 20GB free storage (more for GPU acceleration)
🌐
Outbound access to *.nexusai.cloud (ports 443, 8443)

🚀 Choose Installation Method

Quick global installation using npm or standalone binary. Recommended for developers and local testing.

  1. 1

    Install via npm

    Run the following command to install the CLI globally:

Terminal 
npm install -g @nexusai/cli
nexus init
  1. 2

    Authenticate

    Log in with your API key or OAuth credentials:

Terminal 
nexus auth login
# Paste your API key or open the browser prompt
ℹī¸
On macOS, you may need to run sudo npm install -g @nexusai/cli or use nvm to avoid permission errors.

Containerized runtime with zero host dependencies. Ideal for staging and production.

  1. 1

    Pull the latest image

    Terminal
    docker pull nexusai/engine:latest
  2. 2

    Run the container

    Terminal
    docker run -d \
  -p 8080:8080 \
  -e NEXUS_API_KEY=your_api_key_here \
  -e NEXUS_LOG_LEVEL=info \
  --name nexus-engine \
  nexusai/engine:latest
ℹī¸
For GPU acceleration, add the --gpus all flag and use the nexusai/engine:latest-cuda image tag.

Deploy to Kubernetes using the official Helm chart. Supports auto-scaling, secrets management, and service meshes.

  1. 1

    Add Helm repository

    Terminal
    helm repo add nexusai https://charts.nexusai.cloud
helm repo update
  2. 2

    Deploy with Helm

    Terminal
    helm install nexus-engine nexusai/nexusai-engine \
  --set apiKey=your_secret_key \
  --set replicaCount=3 \
  --set resources.limits.memory="8Gi"
ℹī¸
Production deployments should use Kubernetes Secrets or Vault for API key injection. See our Helm values reference.

Python client for rapid prototyping and ML pipeline integration.

  1. 1

    Install package

    Terminal
    pip install nexusai-sdk
  2. 2

    Initialize in Python

    main.py
    from nexusai import Client

# Initialize client
nexus = Client(api_key="your_api_key_here")

# Load modelmodel = nexus.models.load("nexus-v3-ultra")
result = model.infer("Analyze this dataset...")
print(result)

⚙ī¸ Environment Configuration

Customize your deployment using environment variables or a nexus.config.yaml file.

Variable Default Description
NEXUS_API_KEY unset Authentication token from developer console
NEXUS_LOG_LEVEL info Logging verbosity: debug, info, warn, error
NEXUS_CACHE_DIR ~/.nexusai/cache Local model & token cache directory
NEXUS_MAX_CONCURRENCY 4 Maximum parallel inference requests
NEXUS_GPU_ACCEL false Enable CUDA/Metal hardware acceleration
nexus.config.yaml 
runtime:
  log_level: info
  cache:
    enabled: true
    path: ~/.nexusai/cache
    ttl: 3600

performance:
  max_concurrency: 8
  gpu_acceleration: true
  quantization: int8

api:
  timeout_ms: 15000
  retry_attempts: 3

✅ Verify Installation

Run the following command to confirm your setup is working correctly:

Terminal 
nexus doctor

# Expected output:
✓ CLI Version: 3.2.1
✓ Authentication: Valid
✓ Network: Connected
✓ Disk Space: 24.5 GB available
✓ Runtime Engine: Ready
All checks passed. NexusAI is ready to use.
🎉
Installation complete! Head over to the Quickstart Guide to build your first AI pipeline.

🔧 Troubleshooting

Common Issues

  • ⚠ī¸ Permission Denied (npm global install)

    Use nvm or sudo chown -R $(whoami) /usr/local/lib/node_modules. Never run npm with sudo directly.

  • ⚠ī¸ Docker GPU not detected

    Ensure NVIDIA Container Toolkit is installed. Run nvidia-smi to verify drivers. Use --gpus all flag.

  • ⚠ī¸ API Key Authentication Failed

    Verify the key is copied without trailing spaces. Regenerate from Console if expired.

Still stuck? Check the Full Troubleshooting Guide or contact Support.