This is my personal repo for MLOps Zoomcamp from DataTalksClub

Course repository link

Chapter 1 - Introduction

Environment preparation

Chapter 2 - Experiment Tracking

Create environment - GiHub Codespace

conda create --name exp-tracking-env python=3.9

Activate experiment tracking environment

conda activate exp-tracking-env

Change to directory

cd 02-experiment-tracking

Install requirements.txt

pip install -r requirements.txt

Run MLFlow

mlflow ui --backend-store-uri sqlite:///mlflow.sqlite

Chapter 3 - Orchestration and ML Pipelines

Clone mage repository

git clone https://github.com/mage-ai/mlops.git
cd mlops

Launch mage

./scripts/start.sh

Open mage

Open http://localhost:6789 in your browser.

Chapter 4 - Model Deployment

Deploying models with Flask and Docker

Initialize Environment

pipenv shell
pipenv sync # or pipenv sync --dev

Running App in gunicorn

gunicorn --bind=0.0.0.0:9696 predict:app

Test App

python test.py

Dockerize it

docker build -t ride-duration-prediction-service:v1 .

Run App in Docker

docker run -it --rm -p 9696:9696  ride-duration-prediction-service:v1

Chapter 5 - Model Monitoring

Activate Environment

cd /workspaces/mlops-zoomcamp/05-monitoring/taxi_monitoring
conda activate model_monitoring

Run Grafana Docker

docker compose up --build

Open in browser

Grafana: http://127.0.0.1:3000/login Adminer: http://127.0.0.1:8080/index.php