Over the last decade, AWS has emerged as a leader in the cloud computing space and continues to dominate. According to the latest market report, cloud revenue shot up to $70 billion in 2018 and out of this, AWS cornered the biggest share. Not only that, 2018 seemed to be a golden year as cloud computing spending accounted for more than 20 percent of the total IT budget for organisations.
A market report hints that despite the shift towards multi-cloud deployments, market leader AWS has extended its dominance with about 35 percent of global cloud services by the end of 2018. In fact, such is the dominance of Andy Jassy-led AWS that it has the biggest market share of its four closest rivals (Azure, Google, Alibaba & IBM) combined.
Sign up for your weekly dose of what's up in emerging technology.
Meanwhile, the continual shift to AI has also led to the rise of an intelligent cloud with top tier cloud providers broadening their AI solutions. Analysts reckon AI in the cloud is anticipated to increase at an annualized rate of 50% through 2025. With cloud becoming a critical AI enabler, the market leader has made AI more accessible through its cloud services.
Analytics India Magazine spoke to Madhusudan Shekar, Head of Digital Innovation at Amazon Internet Services Pvt. Ltd. to understand how AWS enables developers to add machine learning to their production applications quickly and how AI and ML is one of AWS’s core strengths. In fact, in the cut-throat cloud services market, there’s been a broad adoption of AWS solutions across the sectors — from manufacturing, financial services, media, entertainment, startups, delivery services, food technology to even agri-tech. Some of the marquee names include L&T, HDFC, SAP, Freshworks and PolicyBazaar among others.
Amongst Cloud Machine Learning Suites Amazon SageMaker Rules
Like other technology majors, AI was a natural progression with Amazon leveraging machine learning technologies to power demand forecasting, product search rankings, warehouse fulfillment centres, inventory forecasting and fraud protection among other use cases. Shekar explains that Amazon has been implementing machine learning at scale for the last two decades starting from the very early recommendation system that was built to recommend books for customers to buy on Amazon portal. “Since then, every aspect of what we do is customer focused, for example, the forecasting we do for our inventory in the warehouses to make sure your order reaches to you in two days. Practically every aspect there is ML-powered, for example, we use computer vision technology to recognize objects and ensure the right items are shipped to you,” he shared.
And as AI became the new normal, every organisation is focusing on gathering accurate data, getting it organised and labelled. But this proves to be a continuous challenge and it’s here that AWS’s cloud-based services automates every each stage of the machine-learning pipeline with the build-train-deploy model with a low investment cost.
Over the last few years, Amazon has doubled down on democratizing machine learning development with platforms like Amazon SageMaker, that allows developers and data scientists to build, train, and deploy machine learning models quickly. A fully-managed service, Sagemaker — covers the entire machine learning workflow — label and prepare your data, choose an algorithm, train the model, optimize and scale it for deployment and make predictions. “Sagemaker is our studio, it encompasses all of the things necessary for you to go ahead and build out a fully managed machine learning capability. So, we started in 2017 by putting together the basic pieces which is the ability to build ML model, the ability to train them and then effectively go ahead and deploy and scale it out. Now, it is the best place to run all forms of AI including TensorFlow,” he said.
Resolving Data Labelling & Multiple Hardware Challenges
The next step was focusing on the cost, the need for speed and ease of use. Also, the real challenge in machine learning building and training path is data labelling. Amazon SageMaker Ground Truth service allows to build accurate training datasets quickly through its chain of public and private human data labelers. Another challenge data scientists face after building and scaling the model is the need to deploy it in different infrastructures. That’s where Sagemaker Neo comes into play — an open source model that allows developers and data scientists to customize the software for different devices and applications. Case in point, if one trains on an Intel CPU, SageMaker Neo optimizes the neural network for specific hardware that the developer wishes to deploy on, like an IoT device. In addition to this, Sagemaker comes with 17 built-in algorithms like Random Forest, Principal Component Analysis among others that can be easily used.
Making AI Accessible Through Learning Programmes
As the clamour for accessible AI and learning gets stronger, cloud computing giant AWS, like other tech majors has launched cloud certifications and formal training programmes which cover the different product features. One of the most popular certifications is Solution Architect which covers the gamut of features, to understand the developer side, operational side and most importantly security side. As part of the mission to put data science and machine learning in the hands of developers, Amazon launched Machine Learning University in November 2018 by open sourcing its machine learning courses used to train engineers at Amazon. As per the announcement, there are more than 30 self-service, self-paced digital courses with more than 45 hours of courses, videos, and labs for four key groups: developers, data scientists, data platform engineers, and business professionals.