Hello, cloud gurus! I’m Scott Pletcher, coming to you from Las Vegas atAWS re:Invent 2021 where I’m hanging out with AWS Senior Developer Advocate Banjo Obayomi. And in this post, we’re going to chat about the big news from re:Invent Wednesday with a round-up of all the machine learning news.
Read on for all the ML and data news fit to post!
Accelerate your career
Get started with ACG and transform your career with courses and real hands-on labs in AWS, Microsoft Azure, Google Cloud, and beyond.
This conversation has been edited for brevity, clarity, and general awesomeness. Any mistakes are probably (OK . . . definitely) on the part of the editor.
DynamoDB Infrequent Access Tables
Scott:The theme of Wednesday was data and machine learning and AI.
One announcement that I thought was really interesting is DynamoDB Infrequent Access Tables. This is something similar to like S3 or EFS infrequently accessed. Do you access it less, you pay less. They had a great example during the keynote, how a social media app can utilize like a photo from five years ago. You’re not going to access it every day, but when you do access that, you want that to come up quickly. So I’m really excited to see what customers can do with this new tier of database.
SageMaker Ground Truth Plus
Scott: Next up, a curious announcement to me was SageMaker Ground Truth Plus.
Banjo: It’s better!
Scott: Yeah, I guess so. Yeah, just add a “Plus” to it.
So SageMaker Ground Truth is a service that customers can use to help clean and classify data before it’s used for machine learning purposes, which is really important.
“Plus” I guess signifies that some, some of that data may require special skills to quantify and classify, and this data will now be routed to people with those special skills. Is that right?
Banjo: Yeah, you need like subject matter experts, SMEs, to really dig deep into these, this data classification problem that maybe a normal person cannot classify. So it’s great to have these experts help with that classification.
SageMaker Serverless Inference
Scott: Fantastic. Another interesting thing, SageMaker Serverless Inference. Now, normally when you deploy inference infrastructure, you have to deploy it on what in essence is EC2 infrastructure, right? This holds the promise to be able to deploy that without us having to do anything.
Banjo: Yeah, I love serverless things because you don’t have to worry about the servers. So be interested to see how we can actually use Inference without having to worry about spinning up an EC2 instance.
Scott: Yeah. One of the questions I do have is what’s the warmup time to, uh, calling these inferences. That’ll.
Banjo: That will be interesting to see once you get hands on and play with that, yes.
Amazon Kendra Experience Builder
Scott: Indeed. Next up: Amazon Kendra Experience Builder. Now of course, Amazon Kendra is a search engine that is designed to be deployed to enterprise organizations to help find documents.
Banjo: Yeah, so with the Experience Builder, you get end-to-end kind of low code, no code workflow that you can use to build your own actual private search, databases, dashboard without having to use custom code. So we’re really excited to see what customers are going to build with this new experience.
Amazon Lex Automated Chatbot Designer
Scott: Big question here. Can we make chatbots less disappointing? That’s a big question for me. And apparently Amazon’s going to try, because they’ve announced Amazon Lex Chatbot Designer. The idea here is that it helps build chatbots and make them better.
Banjo: Chatbots, they’re always a hard learning curve to get through chatbots. Hopefully through this streamlined experience, you can kind of get the intents and see what invokes those different intents to make it much easier to build chatbot experiences instead of having to go through the code all yourself.
SageMaker Studio Lab
Banjo: I’m really excited about SageMaker Studio Lab, because you can spin up a Jupyter Notebook without having to have an account — just an email address and you have a SageMaker deployment ready to go and you can play around and test the machine learning workloads. And when you’re ready, you can migrate that to SageMaker Studio. So I’m really excited that this will help to onboard more customers so they can experience machine learning and build cool things.
Scott: One of the big kind of challenges or hurdles for machine learning right now is just that the skills gap. And by the way, I know a company that can help with that. (Cough, cough — sign up for a free ACG account today, no credit card needed, or nab a Cyber Week discount of 40% off on an annual membership for a limited time — cough, cough.)
The data-driven enterprise
Scott: Right before this filming, I had the opportunity to sit into a session called ENT20 — Data-driven enterprise: Going from vision to value.
And, by the way, this is why I love re:Invent. I intended to go to one session, but I was a little bit late; it was completely filled up. So I just changed course and went to the session across the hall and it was absolutely amazing — especially if you’re at that point in your organization where you’re trying to figure out, “I have data, but how do I turn that data into value? I have this vision, how do I turn that into value?”
I saw cameras there, so they were filming it, and hopefully it ends up online. If you’re in those early stage of trying to figure out “how do we make sense of all that this?” you’ll get a lot out of it.
Scott: Lastly, I just wanted to give a shout out to the DeepRacer team. The DeepRacer finale was held today, and the 2021 DeepRacer championship team is JPMC-Rogue Hyderabad.
Keep up with re:Invent 2021
Well, with that, we’re going to bid you adieu. Stay tuned to all things re:Invent by following ACG on Twitter and Facebook, and subscribe to A Cloud Guru on YouTube. And check out the ACG and Pluralsight re:Invent content hub for even more news and AWS resources. Chat it up with cloudy-minded people on the ACG Discord Community. Keep being awesome, cloud gurus!