MLOps Example

This MLOps example provides a sample AI production setup using Google AI Platform, Tensorflow 2.0, Google Functions(Serverless API), sacred(Experiment Management) and Microsoft NNI(AutoML)

Tutorial

• Korean

https://brunch.co.kr/@chris-song/101

• English(WIP)

http://chris-chris.ai

Install MongoDB

https://docs.mongodb.com/manual/tutorial/install-mongodb-on-os-x/

Install requirements

pip install -r requirements.txt

Download Dataset

$ wget https://archive.ics.uci.edu/ml/machine-learning-databases/wine-quality/winequality-white.csv

Train

GOOGLE_APPLICATION_CREDENTIALS=[crendentials json file] python train.py

GOOGLE_APPLICATION_CREDENTIALS=chris-loves-ai-key.json python train.py

or

export GOOGLE_APPLICATION_CREDENTIALS=chris-loves-ai-key.json
python train.py

Install NNI

pip install nni

Run AutoML Experiment

nnictl create -c nni.yml

Install sacred

pip install sacred

Install omniboard

npm install -g omniboard
omniboard -m hostname:port:database

Local

omniboard -m 127.0.0.1:27017:experiments