Coven of Wisdom - EHV - AWS Sagemaker Workshop
Coven of Wisdom - EHV - AWS Sagemaker Workshop
By Maarten Van Hoof
3 min read
The Coven of Wisdom Eindhoven meetup on Nov. 16th, in partnership with AWS and the amazing Samuel Sopeju from AWS, was a total blast! The goal was simple: dive headfirst into the deep end of machine learning using Amazon SageMaker. And we did just that!
- Authors
- Name
- Maarten Van Hoof
- linkedinMaarten Van Hoof
- twitter@mrtnvh
- Github
- githubvanhoofmaarten
- Website
- websiteBlog
TL;DR
Participants got hands-on experience with Stable Diffusion on AWS, we learned how to conjure a Gen AI system via Amazon SageMaker to create high-res images at scale. Plus, we built a pipeline for auto-tuning models.
Stable Diffusion on AWS
The evening kicked off with a peek at what the Stable Diffusion works in the AWS cloud, using the Stable Diffusion web UI based on Gradio and the AWS Extensions for the Stable Diffusion UI that highly simplifies the integration between the AWS ecosystem and the Stable Diffusion Web UI project.
We played with both positive and negative prompts and got creative with images and ControlNet, a way to reshape and repaint existing images with the help of Generative AI.
Building your own Gen AI REST API
Next up, we started uploading and using different models, and even created our own Generative AI REST API. This involved AWS Lambdas and AWS SageMaker to create a RESTful API that makes generating new content a breeze. We learned to send prompts through request bodies and get an URL with the generated image stored in an S3 bucket in return.
Now, let's scale
To wrap things up, we learned how to scale these solutions. We saw how a Gen AI solution wrapped in an AWS Lambda springs into action every time a new file is uploaded to a specific S3 bucket, uses a Jupyter script as instructions for a Gen AI system and stores the newly generated output in another bucket. This was a great way to see how to apply everything we learned this evening on a bigger scale.
Amazon Bedrock
At the end we even got a sneak peek of Amazon Bedrock, a Gen AI ecosystem that enables us to setup GenAI models, APIs and other AI systems with far less configuration, which has been carefully abstracted away from a developers perspective.
Conclusion
After the workshop all participants got together and enjoyed a delicious Persian meal catered by our good friends at Saffraan. (Yes, I am mentioning the caterer, because oh boy, so good 🤤)
In a nutshell, this AWS Immersion Day workshop was a total win. It was an afternoon full of learning, whether you're an ML pro or just dipping your toes in.