The Opentrons AI application's server.

• This folder is not plugged into the global Make ecosystem. This is intentional, this is a serverless application not tied to the Robot Stack dependencies.

1. clone the repository gh repo clone Opentrons/opentrons
2. cd opentrons/opentrons-ai-server
3. Have pyenv installed per DEV_SETUP.md
4. Use pyenv to install python pyenv install 3.12.3 or latest 3.12.*
5. Have nodejs and yarn installed per DEV_SETUP.md
   1. This allows formatting of of .md and .json files
6. select the python version pyenv local 3.12.3
   1. This will create a .python-version file in this directory
7. select the node version with nvs or nvm currently 18.19*
8. Install pipenv and python dependencies make setup

1. AWS credentials and config
2. docker

python -m pipenv install pytest==8.2.0 --dev

python -m pipenv install openai==1.25.1

• handler
  • the router and request/response handling
• domain
  • business logic
• integration
  • integration with other services

1. Make your changes
2. Fix what can be automatically then lint and unit test like CI will make pre-commit
3. make pre-commit passes
4. run locally make run this runs the FastAPI server directly at localhost:8000
   1. this watches for changes and restarts the server
5. test locally make live-test (ENV=local is the default in the Makefile)
6. use the live client make live-client

• Our first version of this service is a long running POST that may take from 1-3 minutes to complete
• This forces us to use CloudFront(Max 180) + Load Balancer + ECS Fargate FastAPI container
• An AWS service ticket is needed to increase the max CloudFront response time from 60 to 180 seconds