Whether we like it or not, Artificial Intelligence or AI is here to stay. So regardless of what Elon Musk says, we need to embrace the technological changes as they happen.

So to find out more about Southeast Asia’s role in AI and how far off we are from the rest of the world, we spoke to Roy Kim from Pure Storage. Roy is the AI Lead and Director of Products and Solutions for the enterprise cloud data storage firm, and an ideal person to provide us with the insights we need.

Read what Christopher Quek from TRIVE thinks about Singapore’s startup ecosystem

Roy talks about AI in Southeast Asia and where he sees both gaps and great potential for the region. He also shares insights about AI in general, the potential and the path it is on at the moment.

Read what else Roy had to share below.

AIRI Rack.jpg
The AIRI rack. Image courtesy of Pure Storage

Tell us about yourself. What’s your story and how did it lead up to you becoming the product expert for AIRI?

Prior to joining Pure about a year ago, I spent eight years at NVIDIA helping build their GPU computing business. The idea was that serial nature of CPUs was hitting walls related to physics and that massively parallel GPUs would provide performance for applications that need it.

It just happened to be that the path the AI industry was on, and the path that GPU computing was on, crossed around 2012. Since then, I spent most of my time devoted to engaging the AI community with solutions they need to keep pushing forward.

AIRI is inspired by some big gaps in the industry that we wanted to bridge, gaps that only AI practitioners would experience.

Artificial Intelligence is the hot industry right now from here to Silicon Valley. Are there any Asian companies that have caught your eye in the AI space? Anyone from Southeast Asia doing anything interesting?

AI represents a once-in-a-generation opportunity for nations, not only industries, to take lead. It’s because the fundamental technological foundation has shifted with AI. Many Asian countries are investing heavily in AI, including China, Japan, and South Korea. Yet, some of the most cutting-edge AI applications are emerging across Southeast Asia. The region is seeing AI startups across various industries from FSI and e-commerce to less tech-savvy sectors such as agriculture.

For example, the Singapore-based ViSenze delivers intelligent image recognition solutions that shorten the path to action as consumers search and shop online. It works with retailers such as Zalora, Rakuten, and ASOS to convert images into immediate product-search opportunities, improving conversion rates.

Meanwhile, Vietnamese crop intelligence startup Sero encourages rice growers to upload photographs of sick crops online, which are used to train computers how to identify crop diseases. Once the technology has developed further, the company will share crop disease diagnosis and treatment recommendations with farmers via a smartphone app. This helps farmers save money, given that they currently rely on their experience to treat crop diseases, and often spray too many chemicals into their rice fields.

Read which AI companies in Southeast Asia are doing things differently

Recognising the transformational potential AI has across sectors, we worked with NVIDIA to launch AIRI, the industry’s first complete AI-ready infrastructure. AIRI is built for the 99%, those enterprises, and organisations trying to jumpstart their AI initiative. It’s complex and hard to do. AIRI makes it easy.

China recently overtook the US in terms of funding for AI research. What does this mean for the US exactly? Should US companies start to worry or is this natural market progression, where there’s enough to go around for everyone?

AI represents a unique opportunity to advance technologies and societies, so it’s no surprise that nations are investing heavily in AI.

While there’s a tremendous focus on AI, it’s easy to forget that the most important asset in any AI initiative is data.

Organisations with the most data will have a strong position to lead. I’d say that’s where any company, US-based or not, will have to focus.

Coming back to AIRI, what were the reasons behind creating this product? From what we can understand, it looks like a ready-made infrastructure for companies to leverage existing or new AI solutions faster. Is this right? And does this mean that companies haven’t been able to efficiently implement AI into their processes or systems properly?

Everything changes from the old era to the new AI-era. From how you write software to the types of processors you deploy, to the storage system you need, everything changes.

So enterprises are stuck. They’ve been learning one way of building competitive advantage, but much of that experience goes out the window, and they have to re-learn.

For example, while they will continue to need software developers, they will need more investment in data scientist teams.

The same problem is true in infrastructure. Today, a handful of customers have blazed the trail, learning the hard way, through trial-and-error, by building their own infrastructure, on what is the best way to build a system for AI. We felt that this is a huge barrier for the 99% of enterprises. That’s why NVIDIA and Pure joined forces to architect AIRI.

We created AIRI to bring AI-at-scale to all enterprises and shatter the barriers of infrastructure complexities needed for AI. AIRI is an AI-in-a-box solution, combining hardware and software to enable organisations to jumpstart or scale their AI initiatives more easily and quickly. AIRI also supports companies’ ever-increasing data volume and velocity across various workloads today and in the future.

Image may contain: 1 person, sitting
Pure Storage launch event in 2017. Image courtesy of Pure Storage Facebook.

Apparently, AIRI will make data scientists four times more productive. Give us the reasons why and any examples you have would be great.

Today, data scientists often wait a month or more for a single job to finish. Can you imagine that? They literally kick of a job into their compute system, then wait around for a month for it to finish!

So what’s the problem? Well, it’s because the job runs on a single computer. The question we asked is, is it possible to accelerate the job across more computers, all working on the same problem to finish it faster? While this idea has been talked about, it turns out that the process of running a single job across lots of computers is not easy.

That’s the problem we solved with AIRI Scaling Toolkit. AIRI comes with four DGX systems and the toolkit enables data scientists to easily scale a single job across four computers, giving a four-fold increase in performance.

Where do you think there would be the greatest impact if AIRI was widely adopted?

AIRI will benefit any industry or company exploring AI, but particularly so for companies that run data-intensive storage workloads, such as real-time and big data analytics in BFSI, healthcare, and professional sports. For example, Paige.AI, an organisation focused on revolutionising clinical diagnosis and treatment in oncology through the use of AI, requires the most advanced deep learning infrastructure available to quickly turn massive amounts of data into clinically-validated AI applications. The powerful combination of NVIDIA DGX-1 and Pure Storage FlashBlade accelerates Paige.AI’s mission to transform the pathology and diagnostics industry from highly qualitative to a more rigorous, quantitative discipline.

Imag courtesy of Zenuity Facebook

In the automotive industry, Zenuity, a joint venture between Volvo Cars, the premium carmaker, and global airbag manufacturer Autoliv, the worldwide leader in automotive safety systems, selected FlashBlade and NVIDIA DGX-1 as the foundation for its machine learning project to put the safest autonomous vehicles on the road by 2021. Each vehicle is equipped with sensors such as LIDARs and cameras to safely navigate in its surroundings. Millions of frames collected from the cars are used to train deep neural networks to power the software that runs Zenuity’s fleet of self-driving vehicles.

In Southeast Asia, we’re far behind China and US in terms of AI, but there are a few companies venturing into that space. This includes the Singapore government with their SmartNation push, but what’s holding us back from becoming a real player in AI? Is it talent, infrastructure, or did we just start really late?

Singapore is working towards filling an AI talent gap – data scientists are one of the most in-demand professionals today, but the current supply doesn’t meet the ever-increasing demand. Nevertheless, specialisation in AI is gaining ground, fuelled by the Singapore government’s efforts to advance the AI agenda through training and reskilling initiatives, as well as resources to develop data scientists.

Apart from that, big data has now transformed to big and fast data, and they need to be delivered to parallel processors quickly to train AI algorithms faster and more accurately. As large, high-quality datasets are needed to ensure accuracy in deep learning models, data collection and cleaning are critical for AI applications. Storage infrastructure, therefore, needs to be able to support today’s massive data volume and velocity across workloads. Even so, many companies still rely on legacy platforms such as spinning disks, which struggle to meet the demanding requirements of AI.

However, the good news is that there has been tremendous progress recently in the development of technology and software-driven data infrastructure to simplify and accelerate AI deployment. The rise in modern all-flash platforms has enabled companies to balance compute, memory and storage, and optimise AI workloads, by storing data into a centralised hub and enabling various workloads to run simultaneously.


This is a new long-form interview section focused on delving deeper into industry topics and understanding the situation from a ground-up level. If you have a founder or industry expert in mind, which you believe would fit these criteria. Please drop us a message here.