interesting takes on CI-CD (in general):
https://stackify.com/continuous-delivery-vs-continuous-deployment-vs-continuous-integration/
https://medium.com/onfido-tech/continuous-integration-for-ml-projects-e11bc1a4d34f
This is quite a simple/elegant method of CI for ML! have to think about it further.
should i sit my (accuracy) test data in s3 too? hmmm
more options for production level machine learning
https://github.com/ethicalml/awesome-production-machine-learning
it’s clear from here there are many challenges…!
we need to first get our data into shape. which the entire team can use. we’re talking about standardizing the etl process.
Then, from there we can standardise the machine learning process - ci for ml (see link above).
on a bigger level - who do we hand it over to? are there production level ML engineers vs research level (data scientists?)
questions, questions, questions.
what is the ideal world we want?