ML Assignment

Please implement a translation inference service that runs on Kubernetes and provides a RESTful API on port 9527.

The translation model is M2M100, and the example can be found in app/translation_example.py.

You should first fork this repository and then create a PR when you're finished.

Delivery

• app/Dockerfile: To generate an application image
• k8s/deployment.yaml: To deploy image to Kubernetes
• Other necessary code

Input/Output

When you execute this command:

curl --location --request POST 'http://127.0.0.1:9527/translation' \
--header 'Content-Type: application/json' \
--data-raw '{
    "payload": {
        "fromLang": "en",
        "records": [
            {
                "id": "123",
                "text": "Life is like a box of chocolates."
            }
        ],
        "toLang": "ja"
    }
}'

Should return:

{
   "result":[
      {
         "id":"123",
         "text":"人生はチョコレートの箱のようなものだ。"
      }
   ]
}

Bonus points

• Clean code
• Scalable architecture
• Good inference performance
• Efficient CPU/GPU utilization