API-based deployment in Prompt Engineering / GenAI - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is API-based deployment in machine learning?

API-based deployment means making a machine learning model available through an Application Programming Interface (API) so other programs or users can send data and get predictions easily.

beginner

Why use an API for deploying machine learning models?

Using an API allows many users or applications to access the model remotely, making it easy to integrate predictions into websites, apps, or other software without sharing the model code.

intermediate

Name a common protocol used for API-based deployment.

HTTP (HyperText Transfer Protocol) is commonly used, often with REST (Representational State Transfer) style APIs to send requests and receive responses from the model server.

beginner

What is a typical input and output in an API-based ML model deployment?

Input is usually data in JSON format sent in a request, like text or numbers. Output is the model's prediction or result, also in JSON, sent back in the response.

intermediate

How do you ensure your API-based ML deployment is scalable?

You can use cloud services, load balancers, and container orchestration tools to handle many requests at once and keep the service fast and reliable.

What does API stand for in API-based deployment?

A Artificial Processing Input

B Automated Prediction Integration

C Application Programming Interface

D Advanced Programming Interaction

API stands for Application Programming Interface, which allows different software to communicate.

Which data format is commonly used to send input data to an ML model via API?

A XML

B JSON

C CSV

D YAML

JSON is widely used because it is easy to read and write for both humans and machines.

What is a key benefit of deploying ML models via API?

A The model runs faster on local machines

B It requires no internet connection

C Only one user can access the model

D Multiple applications can use the model remotely

APIs allow many applications to access the model remotely, enabling easy integration.

Which protocol is most often used for API communication in ML deployment?

A HTTP

B SMTP

C SSH

D FTP

HTTP is the standard protocol for web communication and APIs.

How can you handle many API requests to your ML model at the same time?

A Use cloud services and load balancing

B Ignore extra requests

C Use a single powerful computer only

D Limit the model to one request per hour

Cloud services and load balancers help distribute requests to keep the service responsive.

Explain how API-based deployment makes machine learning models accessible to other applications.

Think about how apps talk to each other over the internet.

Describe the steps to deploy a machine learning model using an API.

Consider what happens from model ready to user getting predictions.