Amazon re:Invent 2019 Las Vegas
I’m not in Las Vegas(I wish to be there) but watching a live keynote session by Andy Jassy, CEO of AWS. Huge updates on AWS Machine Learning services, SageMaker Studio, SageMaker AutoPilot, CodeGuru, and AWS Wavelength.
Really important question:
What’re the next set of machine learning challenges?
SageMaker Studio is the first fully integrated development environment for Machine Learning.
SageMaker Autopilot is an automated training platform which read training data from CSV file (via S3), automatically transform data with dates, days, address, etc. Based on the data platform, selects the best algorithm, either classification or regression. Users can define different models, and they support up to 50 different models and configurations. View models and compare accuracy in a notebook. Once a user selects a model automatically deploy in a single click. I would love to work on the SageMaker Autopilot and compare the manual and automated model outputs.
For Developers, AWS included new services for a developer called Amazon CodeGuru. CodeGuru can pull requests from Github and CodeCommit(AWS). Then, based on reviews from Amazon and the top 10,000, open-source projects do code reviews and provides comments on pull requests. More important is that it can identify lines of code, issues, recommends remediation, and links to relevant documents. I’m going to integrate into my company and see the code review and recommends.
In the end, he hammers out three main breakings through barries as AWS Outposts, AWS Local Zones, and AWS Wavelength (Storage at the edge of 5G network and single-digit millisecond latency to mobile devices).
