OpenAI with Java

The project demonstrate the integration with Azure OpenAI text-davinci-003 model, Azure Storage Queue using JAVA. The purpose of this repository is to provide a reference point with the Azure OpenAI "text-davinci-003" model serving as an example POC to build a java WebJob or for an independent local run test.

Demo App Architecture

Pre-requisite

• Azure Open AI service created and text-davinci-003 model deployed
• Have a Azure Storage source and destination queues created
• Acquire the below config information and have the needed environment variables setup

How to Run

• Install platform specific Java 17 onwards, e.g. for windows https://learn.microsoft.com/en-us/java/openjdk/download
• Make sure JAVA_HOME environment variable is configured
• Install mvn(maven) https://maven.apache.org/download.cgi

Build

• Clone the repo locally
• Navigate to the cloned directory
• run "mvn compile" then "mvn package"
• Running "mvn package" would generate the jar file "openapi-javaapp-1.0.0.jar" under \target directory

Running jar locally

• Have these environment variable configured that contains the configuration for the azure open AI, azure storage queue
  • AZURE_STORAGE_CONNECTION_STRING
  • AZURE_STORAGE_SOURCE_QUEUE_NAME
  • AZURE_STORAGE_DESTINATION_QUEUE_NAME
  • AZURE_OPEN_API_KEY
  • AZURE_OPEN_API_ENDPOINT
  • AZURE_OPEN_API_MODEL_DEPLOYMENT_NAME
• Run the jar using "java -jar target\openapi-javaapp-1.0.0.jar", if there's no error and run completes and exits, then the jar content is all good to be deployed as webjob in azure.
• To see the content of the jar file run "jar -tf target\openapi-javaapp-1.0.0.jar" or extract the jar content locally

Running As WebJob in Azure

Currently EastUSEUAP2 has the code deployed that enables running webjobs on Linux.

• Create a Java Linux container webapp, please make sure the container, custom or blessed image based have Java 17 + in it, JAVA_HOME environment variables configured, and WEBSITES_ENABLE_APP_SERVICE_STORAGE environment variable set to true for the web app
• Have the above mentioned environment varialbes configured as app settings
• Please zip these files "openai-java-webjob-start.sh", "settings.job" and "openapi-javaapp-1.0.0.jar". And deploy the zip file as continuous webjob, either by zip deployment or by calling webjob PUT API, details here https://github.com/projectkudu/kudu/wiki/WebJobs-API