This is a small Flask project that lets you run NLP on AWS Lambda hosting Spacy models on S3 and therefore not running into the 250MB Lambda limit.

pipenv install

pipenv shell

pipenv install -d to install also dev dependencies
pytest

Create an .env file in the root folder where settings.py is with these variables

FLASK_ENV=development
SENTRY_DSN=your_sentry_dsn
SPACY_MODEL_MEDIUM=en_core_web_md-2.3.1 # or whatever you want
SPACY_MODEL_SMALL=en_core_web_sm-2.3.1 # same
S3_BUCKET=your_bucket

Delete zappa_settings.json and create one from scratch if you want, or change the S3 bucket name in the JSON provided. I used different buckets to store the models and the lambda code but you can probably use the same one.

Then, do zappa deploy dev and it should be deployed in a couple minutes