Field Notes from re:Invent 2018 – AWS is All In on Machine Learning and AI

Field Notes from re:Invent 2018 – AWS is All In on Machine Learning and AI

While most people stayed home for Thanksgiving or battled Black Friday crowds, engineers from around the world descended on Las Vegas for AWS re:Invent 2018. The conference more affectionately referred to simply as “Reinvent” is Amazon’s annual cloud extravaganza. For the fifty thousand plus who attended, occupying nearly every hotel on the strip, Amazon provided a broad vision of the future through a frenetic series of product announcements, bootcamps, workshops, and focused technology summits.

AI has become so important to the future of AWS, that a separate AI Summit was carved out to go through real-world use cases and deployments. It was truly enlightening to see how AI has been used in areas such as disease prediction in medicine, traffic engineering, all the way to tax preparation.

A Perfect Event for Prophix

As a SaaS provider, AWS has been a valuable technology partner to Prophix and this was my second year attending Reinvent. AWS certainly likes to run their conferences differently than others, mixing key events throughout the day and night, such as Midnight Madness and Monday Night Live—a full blown 90-minute prime time keynote, complete with beer!

The announcements throughout the week present so many innovation opportunities for Prophix. We have already invested heavily in the areas of AI whether it’s natural language processing, natural language generation, anomaly detection, and machine learning. There is no shortage of creative ideas flowing through the team.

Pervasive Machine Learning and AI

I was most interested in the evolution of machine learning technologies this year, and AWS did not disappoint. At Reinvent 2017, AWS released SageMaker which provided access to cloud scale machine learning services to developers, and this year, they added even more to their AI and ML services platform. Even more impressive is the fact SageMaker eclipsed 10,000 customers and statistics showing that the bulk of all global cloud machine learning workloads is running on AWS.

To further support the many machine learning workloads, Amazon launched Timestream, a new cloud database dedicated to capturing and processing time series data, probably the most pervasive data set in the business world when you think about corporate performance metrics, financial data, and also data from IoT devices. While Amazon SageMaker Ground Truth is a new complementary service to significantly streamline the process of labeling data sets to be used for training machine learning models.

Machine Learning is an area that has the potential to significantly enhance FP&A processes, through the incorporation of myriad data points now being collected by businesses. Coupled with exogenous macro level data of key events, economic indicators, even weather patterns, the possibilities of using machine learning algorithms to enhance forecasting is compelling.

The most exciting news is AWS’ announcement of launching a fully managed deep learning time-series forecasting service, preloaded with algorithms already utilized by Amazon.com for business and operational forecasting. Imagine the experience Amazon’s own FP&A and operational groups have had forecasting at scale! This is a game changer. It allows customers to gain access to not only a highly scalable forecasting engine, but one that can also learn from the data supplied, automatically choosing the best algorithms for training a model. AWS demonstrated that in many cases, forecasts created with the benefit of machine learning outperforms regular forecasts done purely using traditional statistical methods by as much as 50%.

Finally, a presentation I found very interesting was on Quantum Computing and the potential game changing impact on machine learning. While still in its infancy, quantum computing machine learning could solve many optimization problems seen in the business world, with many variables and analysts interested in multiple outcomes, at blazing speeds.

Enter the AWS Hybrid Cloud

One somewhat surprising announcement at Reinvent is the launch of a new hybrid cloud offering. AWS Outposts gives companies the ability to run AWS cloud services using the same set of hardware utilized by AWS in their own data centers. AWS will ship a data center ready rack containing compute and storage resources as part of the Outposts offering.

What’s important to note is this is not simply a private cloud, Outposts is a logical extension of a customer’s existing pool of AWS resources in the nearest region. Customer’s I spoke to were extremely excited at the possibilities of creating a true hybrid cloud to fit the various workloads to satisfy corporate requirements.

Prophix has always been a defender of choice for consumers when it comes to deciding on application consumption and deployments between cloud or on-premise, encouraging decisions to be based on overall business strategy.

Like many of the attendees, I am also excited by this announcement as it not only provides choice but will also be valuable in the future for addressing data access needs for machine learning applications. Let’s face it, companies will continue to have myriad data sources located on-premise and in the cloud, AWS Outposts will permit access to all corporate data using the same set of resources available in the full AWS stack. This is fantastic for ISVs like Prophix that have already invested heavily in integrating with the AWS cloud.

The AWS Racing League?

That’s right a racing league, but with a twist:  Amazon launched their DeepRacer self-driving reinforcement learning scale race car, complete with its own racing league against other AWS developers!  I suspect this announcement was more to add a fun factor for cloud and AI engineers.

These 1/18th scale radio-controlled, four-wheel drive pocket sized racers feature an Intel Atom CPU, built-in cameras, wifi, and are fully connected to the AWS AI cloud infrastructure. Developers train the cars to race on various tracks by using reinforcement learning algorithms. I think it’s a brilliant way to provide some practical and fun applications for engineers to learn about machine learning and cloud. I need to order a few Deep Racers for Christmas!

Final Thoughts

AWS Re:Invent is a perfectly timed event to wrap up a busy year. The pace of innovation at AWS is mind boggling. One attendee I spoke to noted that nearly 1800 new capabilities were added to the AWS cloud in 2018.  Entire teams are needed just to keep pace with the change. For engineers and innovators, AWS is the perfect partner. I liken their platform to Lego: AWS provides all the pieces necessary to build cloud applications at scale—but it’s up to the builder to innovate. The new AWS capabilities have great potential and will allow Prophix to further realize its vision of the future of finance. I can’t wait for re:Invent 2019!

Geoff Ng

As Vice-President of Product Planning, Geoffrey is responsible for managing the long-term product and technology strategy at Prophix. Geoffrey began his tenure at Prophix as a management consultant leading the implementation of financial analytics systems at over 70 organizations throughout North America.

Download the Prophix CPM Overview
CPM Overview

Let’s Stay Connected!

Sign up to get the latest Prophix blog

Archives