Certainly data scientists are in high demand these days. Businesses can hardly make a move without decoding the dense pattern of metrics that determine how they’re doing, and where the market (might) be heading.
But will the data scientist always be so supreme? Automation – driven by the cloud – can handle an ever growing number of tasks that are now the exclusive province of the data scientist. Tolga Tarhan, CTO, Rackspace, discussed how companies are learning that, as analytics applications grow ever more advanced, regular humans can use them – or at lease regular humans with software development skills.
Cloud Storage and Backup Benefits
Protecting your company’s data is critical. Cloud storage with automated backup is scalable, flexible and provides peace of mind. Cobalt Iron’s enterprise-grade backup and recovery solution is known for its hands-free automation and reliability, at a lower cost. Cloud backup that just works.
Automation handles some (or many) of the processes of machine learning. Next-gen applications like AWS’s SageMaker Studio facilitate this by offering a cloud-based, intuitive visual interface that democratizes the act of data mining.
Does this spell the end of the data scientist? Will cloud automation revolutionize data analytics?
Download the podcast:
SEE TOP QUOTES:
Cloud and the Managed Service Provider
“Managed services is probably one word, but really, it’s broader sort of IT services, helping customers adopt the cloud, and there’s a lot of transformation that goes into a good cloud adoption. Then once you get there, helping you operate in the new environment. And that can be a combination of us doing it for you, us doing it together, or us enabling your team to do it.
“I think the traditional model of MSP doesn’t translate well to the cloud. In fact, our own portfolio has evolved because of this reality. And what customers want now is more of an extension of their team. So whereas maybe 20 years ago, you wanted to outsource the whole problem and just go, “Look, your SLA to me says these apps will be up and running, and that’s that.”
“Today, cloud adoption and infrastructure is so intertwined with application development that you can’t draw this neat boundary anymore that says, “Look, we build the app and you run the app.” It tends to be much more cyclical than that and much more integrated than that. And as a result, what you really want is to get ops experts on the team that’s doing the building and the deployment and the management. And so what we do is supplement your team with that operations expertise. We take out some of the really heavy burden of the 24/7 monitoring and response in the middle of the night. But at the end of the day, we help advance your products and services and solutions to your customers every day, and help our customers innovate faster. So that’s a really different mindset than the old days of, “Look, we just contract to keep the service up and running.”
“One of the things I like to really say is you could have the best team in the world internally working on this, but they’ve only worked on it at your current enterprise, or maybe one or two before that; versus a team of people who have done this every day across hundreds and thousands of customers. So the goal is to avoid the pitfalls and take you right to the end more quickly.”
Cloud-based Automation and Data Analytics
“There was a period in the ’90s that was the AI winter, where AI took a really big step backwards. We didn’t advance substantially for a long time, at least not at the sort of commercial level. And then in the last 20 years, probably even in the last 10 years, it’s advanced very, very quickly. And largely led by very sophisticated data scientists that have studied this, that this is their career is to do this well, to build these models. I think where we’re going… And I think we’re clearly passed, what I would call AI winter. And the AI renaissance is done now, too. And we’re now going into the industrial age, is how I would say it.
“We’re now taking this very powerful technology and making it accessible to all the software developers in the world, where really, the target here isn’t necessarily every person can go do AI. The target is every software engineer can do AI. And that’s a much bigger body of people that now have access to this technology, and that are already working on business problems.
They’re already connected to, “Hey, here’s some requirements or some problem I’m solving for. Here’s some tech I’m building.” If I can integrate AI into that technology the way I can integrate the web into my solutions, or the way I could integrate any other sort of modern tech into that solution, then all of a sudden, the addressable market, the people that can use AI just grows exponentially, and I think the adoption therefore grows.
“So if I could take a low-code solution, combine it with some AutoML… You know again, it sounds crazy when I say it out loud right now, but we’re very close to those things coming together. And I think there will always be a place for the most advanced data scientists, just like there was prior to AI being such a big concept. But the key here is to make the core technology available to more people.”
The Future of Cloud, AI and Automation
“So outside the AI space, I think automation in the cloud is here and now. We’re not waiting for anything. We’re really just waiting for people to adopt, right? You can build environments today that go from code at a source repository, to fully deployed and running with no humans. We do it every day. I think that’s here and now. In the machine learning, in the AIML space, I think we’re gonna see these AutoML tools advance substantially over the next couple of years. And I would be very surprised if all of the hyperscalers didn’t have a series of releases in the space over the next couple of years.
“You look at any sort of major enterprise customer today, they’re definitely… They’ve got people working on ML problems. But how often are those models making it to production? So, there’s the building of a model and the research that goes into it, then there’s deploying it to production and having it do inference. That is, having it actually do its predictions in the wild, in production. And I think that is a very low percentage. I think, unfortunately, a lot of ML project’s stuck as a science experiment. Actually, one of the things I advise customers as they embark on an ML journey is, “Don’t seek the perfect model that has the perfect answer. Actually build something quickly, quickly get it integrated into your production process.”
