From Data to Deployed Model: Build, Train and Fine-Tune Machine Learning Models in the Cloud

About

A model trained on a laptop is a good start. Turning it into something your lab can run at scale and share with collaborators usually takes a lot more work. This hands-on workshop walks through the whole machine learning process on AWS, from raw data to a live model that returns predictions on demand.

You'll work in a familiar notebook environment with a research dataset. First you'll prepare the data, then train a neural network on cloud computing resources and deploy it so it answers new requests in real time. You'll also fine-tune a pre-trained AI language model on your own research vocabulary, which is one of the fastest ways to get specialized results out of today's foundation models. We'll finish with practical habits for keeping cloud costs under control.

Learning outcomes: - Understand the full machine learning lifecycle, from data preparation to deployment - Prepare research data for training and avoid common mistakes that skew results - Train models on cloud computing resources without managing any hardware - Deploy a trained model so it returns predictions in real time - Fine-tune a pre-trained foundation model for a research-specific task - Manage and clean up cloud resources to keep costs predictable

This event is not being recorded. Any materials provided by the presenter will be sent to all registrants by the end of the second business day after the conclusion of the event. Please email ResearchITEvents@ucf.edu to request access to the materials two business days after the completion of the event if you did not register.

Presented by Gabriel Brackman, senior solutions architect, Amazon Web Services