Hands-On OpenAI API: Building a Real-World Solution

Learn how to build an entire AI application that solves real-world problems using OpenAI APIs.

Introduction

• Building a real-world AI app
• What you need to know
• What you are going to build throughout this course
• Preparing the tools and development environment

1. Using the Audio API to Transcribe Audio and the Chat Completions API to Create Documents

• Extracting rich information from audio messages
• Installing the required library and preparing the code
• Transcribing the audio files by using the Audio API
• Creating the JSON document using the Chat Completions API

2. Building the KinderLogger Assistant with the Assistants API

• Bringing the KinderLogger assistant idea to life
• Creating the assistant using the Assitants API
• Creating a new conversation thread and adding a message
• Running and testing the conversation thread

3. Implementing a Retrieval Augmented Generation (RAG) Process in the Assistant

• Using private documents and data in the KinderLogger project
• Uploading the transcription files to the assistant
• Improving the assistant with prompt engineering techniques
• Using the assistant on the Assistants Playground

4. Implementing the Web API to Expose the Assistant

• Exposing the assistant as a service
• Creating the Web API using the FastAPI library
• Calling the assistant from the Web API
• Implementing logging in the Web API
• Building and publishing the Docker container image
• Deploying the Azure Container App for the service

5. Identifying Harmful Content Using the Moderations API

• Applying responsible AI practices in KinderLogger
• Implementing the Moderations API in the transcription app
• Implementing the Moderations API in the Web API
• Building a new image and updating the Azure Container App

6. Creating the KinderLogger GPT

• Using a GPT as a frontend for the KinderLogger project
• Creating the GPT and the action for invoking the service
• Testing the KinderLogger GPT

Conclusion

• Next steps