This repository contains an example application to illustrate how one could perform AI inferencing using DeepSeek. For the sake of this application, DeepSeek has been deployed to a Linode with dedicated GPU (NVIDIA RTX 4000 Ada).
Ollama has been deployed to the Linode machine and the latest DeepSeek model (deepseek-r1:latest) must be pulled. Additionally, the Ollama server has been exposed for responding to incoming HTTP calls.
This application is designed to generate personalized product descriptions, given some information about the potential customer and the general product description.
The Spin application consists of two major components (frontend and api). The frontend component serves a simple HTML5 application at application root (/), which users could use to perform AI inferencing. The api component is implemented using TypeScript and exposes the /personalize endpoint, which responds to POST requests.
To compile the application on your machine, you must have the following tools installed:
To compile the Spin application, run the spin build command from within the root directory of this repository (where spin.toml) is located.
spin buildThe application has the following variables defined, which must be specified when running the app locally or deploying it to a remote runtime:
| Variable Name | Environment Variable Name | Default Value | Required | Description |
|---|---|---|---|---|
ollama_base_url |
SPIN_VARIABLE_OLLAMA_BASE_URL |
none | yes | Ollama API endpoint e.g. http://1.2.3.4:11434 |
ollama_model |
SPIN_VARIABLE_OLLAMA_MODEL |
none | yes | Name of the model you want to use e.g. deepseek-r1:latest |
temperature |
SPIN_VARIABLE_TEMPERATURE |
"0.4" |
no | Temperature forwarded to the LLM upon inferencing |
When running the Spin application on your local machine, you can specify application variables by setting the corresponding environment variables (see table above):
SPIN_VARIABLE_OLLAMA_BASE_URL="http://1.2.3.4:11434" SPIN_VARIABLE_OLLAMA_MODEL="deepseek-r1:latest" spin up