CDK Infrastructure Configuration

• bin/cohere-embed-v3-services.ts: Entry point for CDK deployment. It reads environment variables, checks them, and initializes the CDK stack for deployment.
• lib/CohereEmbedV3ServicesStackProps.ts and related files in lib/constructs: Define AWS infrastructure as code, including VPC, ECR, and App Runner configurations.
• package.json and package-lock.json: Define Node.js project dependencies, including AWS CDK libraries, TypeScript, and Jest for testing.
• tsconfig.json: TypeScript configuration file.

FastAPI Application

• Located in src/coreservices/cohere-embed-v3, it includes a FastAPI application main.py that exposes endpoints for embedding text with Cohere's API.
• Dockerfile and docker-compose.yml: Docker configurations for building and running the FastAPI application. Run on your local machine with command docker compose up --build -d and then checkout via: localhost:8090/docs
• requirements.txt: Python dependencies for the FastAPI application, including FastAPI, Uvicorn, and Cohere's SDK.

Configuring Your Local Environment

You can deploy the stack from the following repository: AWS VPC Creator. After deployment, you can use the VPC_ID obtained from there as an input parameter.

Before running the application locally, you need to configure your environment variables:

1. Copy the .env.example file to a new file named .env in the same directory.
2. Open the .env file and fill in the values for the environment variables as per your setup. This includes setting up the COHERE_API_KEY, DATA_INGESTION_API_KEY, and any other variables you might need to customize for your local development environment.
3. Ensure the .env file is correctly referenced in your application's configuration to load these environment variables.

Demo

Local Demo with Docker

To see a demo of the service running locally with Docker, refer to the screenshot below:

AWS App Runner Demo

After deploying the service on AWS with App Runner, you can see the demo as shown in the screenshot below: