Once the cloud infrastructure or on-premises server has been configured according to the prerequisites, you can install all EVA components using Helm.

EVA is divided into EVA App, EVA Agent, and EVA Vision according to service role. Each component is provided as an independent Helm chart.

Follow the installation process below in order to ensure that the system starts correctly.

🔔 To proceed with the installation, you must have an AWS Access Key for accessing and authenticating with the Mellerikat cloud image registry (ECR).

🛠️ Installation Process

Because EVA components have dependencies, we recommend installing them in the following order.

1. Install dependencies: First, deploy the base infrastructure required by EVA Agent—the Qdrant database and the vLLM inference server.
2. Install EVA Agent: After the dependencies are ready, install the agent responsible for AI inference and data processing.
3. Install EVA Vision: Install the module responsible for vision analysis of video streams.
4. Install EVA App: Install the web interface and administration dashboard used by end users.

🔄 Maintenance and Removal

Helm makes it easy to manage the EVA lifecycle.

• Update: Change the configuration or version in values.yaml, then apply it with helm upgrade.
• Remove: Remove deployed resources with helm uninstall.

(Note: PVCs/PVs created for data persistence may need to be removed manually.)

Choose an Installation Method

• Automated installation: Use EVA Deployer to deploy the infrastructure and EVA components together with Ansible.