hands-on-train-and-deploy-ml

Why we included this project

Most ML tutorials stop at the notebook. This one takes you the rest of the way: you train a model to predict crypto prices using historical data from Coinbase, track your experiments with CometML, then deploy the winner as a serverless REST API on Cerebrium. On top of that, it shows how to set up automated retraining and safe redeploys with GitHub Actions and a model registry, so the pipeline doesn't fall apart when new data arrives. The steps are written as short, runnable lectures with make targets, so it's easy to follow along in bite-sized chunks even if you only have an hour. If you're trying to get a feel for how real MLOps tools fit together without standing up your own infrastructure, this is a practical starting point.