Deploy API to AI Logic Deployer
This document explains step-by-step how to register an LLM-based API developed in a local environment to the platform where ALD and AIC are installed and deploy it to the cloud infrastructure, focusing on CLI commands. Each step covers the necessary commands and practical precautions.
Service API Registration Step
This is the step to register the service API created in the local environment to the platform where ALD and AIC are installed.
Service APIs are managed on a workspace basis.
Account Login (alm login)
First, you need to log in to the AI Conductor installed on the platform where ALD and AIC are installed to be authenticated. To do this, run the following command in the CLI:
alm login
- After running the command, you will be prompted to enter the credentials (email/username and password) of your platform account. Please enter the correct credential information as instructed.
- If the login is successful, an authentication token will be saved, and the authentication will be maintained for subsequent command executions. (You can confirm this with a login success message or user information output)
- Tip: If the login state is not maintained in the CLI environment or the token has expired, run
alm loginagain to re-authenticate. The login session is valid for 3 hours.
When the login is successful, the following log is displayed.
- If you have access to only one workspace in AI Conductor, that workspace is set as the default workspace. If you have access to multiple workspaces, you must set the workspace to access by default.
alm login
Please enter your AI Conductor ID: gildong.hong
Please enter your AI Conductor password: **********
File /home/jeongjun.park/.token/key.json has been removed for initialization.
access_token has been updated in /home/jeongjun.park/.token/key.json
Login success
test-ws has been updated in /home/jeongjun.park/.token/key.json
You can access these workspaces: ['test2-ws', 'test-ws', 'test1-ws']
Please input default workspace: test1-ws
default_ws has been updated in /home/jeongjun.park/.token/key.json
Default workspace: test1-ws
Service API Registration (alm register)
In this step, you register the LLM API service developed locally as a service in the AI Conductor of the platform where ALD and AIC are installed. Use the alm register command and provide the service name and detailed description information. The registration procedure is as follows:
Constraints
- Uppercase English letters, spaces, special characters, and Korean are not supported, and the length is limited to 50 characters.
- The Service API name can only contain lowercase English letters, dashes (-), and numbers.
- If a service with the same name is already registered, you must update it or delete it and re-register it.
alm register --name (Service API Name)
alm register -n (Service API Name)
(Optional) You can modify the image used to create the Service API.
- If you use
--imageas the default, thepython-3.12image is used.
# image )
alm register --name (Service API Name) --image python-3.13-slim
# API )
alm register --name survey-genius-cloud-service
- When the prompt for the service name (Name) appears, enter a unique name that represents the service to be deployed.
- Specify the name concisely and clearly by combining Korean or English letters, numbers, etc. (e.g., my-llm-service).
- When you complete the input, the CLI performs the service registration and, if successful, outputs a unique ID for the service or a success message.
- Caution: Service names cannot be duplicated within the platform. If a service with the same name already exists, registration will be rejected. In this case, try a different name or, if necessary, delete the existing service.
- After registration is complete, the service is internally registered in the registry, and preparations for cloud deployment are complete.
Caution
- When registering via the ALO-LLM CLI, all folders and files in the current location are compressed and registered.
- In this process, the
.envfile is excluded from compression. (It will be delivered later in the activate step.) - When registering, please modify the port in
config.yamlto 80 and register!! If it is not 80, the error below will occur. (Only if you wrote the logic using ALO-LLM) Error: Please change the port number to 80 and register again!
If successfully registered, a log like the one below will be displayed, and you can check the Service API in the AI Conductor installed in the environment where ALD and AIC are installed.
Registration Successful!
Name: survey-genius-cloud-service
Creator: jeongjun.park
Created At: 2025-06-02T08:01:10
Versions: 1
Update Registered Service API (alm register --update)
To update a registered Service API, you can update it with the command below.
alm register --update (Service API Name)
alm register -u (Service API Name)
If the update is successful, you can see that the version increases as follows.
Registration Successful!
Name: survey-genius-cloud-service
Creator: jeongjun.park
Created At: 2025-07-14T07:41:31
Versions: 2
AI Pack Installation and Service API Deployment Step
This is the step to install the AI Pack and deploy the service API registered on the platform where ALD and AIC are installed to the infrastructure environment where the AI Pack is installed. Only one AI Pack with the same name can be deployed and activated, regardless of the version. If you want to change the version, you must deactivate and delete before proceeding. The AI Pack contains the resources necessary for the service API to perform stable operation services.
Install AI Pack and Deploy Service API to the Cloud (alm deploy)
Install the AI Pack necessary for the registered service API to perform stable operation services and deploy the service API. Use the alm deploy command and specify the name or ID of the target service API. An example of executing the deployment command is as follows.
alm deploy --name (Service API Name)
alm deploy -n (Service API Name)
- Immediately after executing the command, the AI Pack installation and service API deployment process begins, and a message that the deploy request has been received and the service access URL are provided in the CLI output.
- If you want to deploy a specific service API version, you can use the
--version (version)or-v (version)option. If you do not enter the CLI version option, it will be deployed with the latest updated version by default.
# API deploy )
alm deploy -n (Service API Name) --version 2
alm deploy -n (Service API Name) -v 2
- After executing the
deploycommand, a log like the one below is displayed.
Deploy Success:
name: (Service API Name)
creator: jeongjun.park
created_at: 2025-07-14T07:41:54
updator: jeongjun.park
updated_at: 2025-07-14T07:41:54
Note
- If the AI Pack you are deploying requires a lot of resources, you can add the
--namespace Trueoption to the CLI to assign a separate namespace. If you do not add the option, it will be deployed to the public namespace created for each AI Conductor workspace. alm deploy --name survey-genius-cloud-service --version 1 --namespace True
AI Pack Activate Step
This is the step to activate the installed AI Pack. Through this step, the deployed service API is turned ON and becomes available for service.
Note
- It reads the
.envfile in the path where the Activate CLI is entered and passes it as an environment variable. - If you want to update the
.envfile, you can proceed in the order ofdeactivate-> update.envfile ->activate.
AI Pack Activate (alm activate)
Use the alm activate command to activate the AI Pack installation and deployed service API. This makes the service API available for service operation.
alm activate --name (Service API Name)
alm activate -n (Service API Name)
A question to determine the spec of the resource to be used will appear as follows. (If there is only one resource, it is automatically selected.)
| Available specs: | ||||||
|---|---|---|---|---|---|---|
| Name | Instance | Instance Type | vCPU | RAM(GB) | GPU | GPU RAM(GB) |
| standard | t3.medium | CPU | 2 | 4 | 0 | 0 |
| high | t3.large | CPU | 2 | 8 | 0 | 0 |
| low | t3.small | CPU | 2 | 2 | 0 | 0 |
| Please select one of the following spec names: high You selected: high |
Activate Success:
name: survey-genius-cloud-service-20250714074406663
creator: jeongjun.park
created_at: 2025-07-14T07:44:06
updator: jeongjun.park
updated_at: 2025-07-14T07:44:06
aipack_activate_url: https://ald-ym.11m-dev.try-mellerikat.com/survey-genius-cloud-service
file saved at: /home/jeongjun.park/jj_0817/test_11/new_chatbot/example_chatbot/.workspace/survey_analysis
/activate_info.json
- The URL above points to the endpoint of the deployed service. Once the deployment is complete, you can access the API through that address.
- You can also test the API or view the documentation through the Swagger UI. In most web frameworks, the Swagger UI (interactive API documentation) is provided by default at the
/docspath. - Example: To access the Swagger UI of the example URL above, open it in your browser by adding
/docs/api/v1like this:https://my11mservice-123456.cloud.example.com/docs/api/v1. On that page, you can check the list of API endpoints and input/output schemas and call them directly. - Note: Immediately after deployment, it may take about 5 minutes for the service to fully start up. You may not be able to access the URL right away.