IT Life: Making Online Adverts Personal
Aaron McKee of Struq needs the cloud and Big Data to serve people ads they can use
Struq personalises adverts online. It says its mission is to transform adverts from a “generic irrelevant communication”, to a “relevant personalised communication”: exxentially, when you visit a site that it’s working with, you should see adverts that actually match your interests.
As you might expect, that takes a fair amount of technology – processing as much information as quickly as possible, to get the best and most useful match back, in the very short attention time of a person browsing the web.
This is just the kind of thing Aaron McKee, chief technology officer of Struq, likes. Since he entered IT in 1994, he has spent most of his career looking at building systems which, in his words, “make interesting decisions quickly”.
What’s been yourfavourite project in your work so far?
At Struq, we’ve focused on applying big data, machine learning principles to ad personalisation and ad serving. Our systems process several billion ad requests a day, which we have to act upon consistently within about 30ms. For every ad request, we have to decide the optimal combination of creative elements to render, whether we think the user is likely to interact with the ad, and ultimately how much we wish to pay to show that ad.
As a startup, we have to be able to accomplish all of this with more brain than brawn (or cash). By far, the pinnacle of my career has been to architect systems that can handle such silly levels of requests, in previously unthinkable times, whilst making exceptionally accurate predictions, with a relatively small number of servers. For a geek, this has been heaven and has leveraged everything I’ve spent my career learning.
What tech were you involved with ten years ago?
A decade ago, I was helping to build web load balancing software for Linux. It’s amazing to see how far technology has come, when commodity servers can now easily handle 10s or 100s of thousands of requests a second.
What tech do you expect to be using in ten years’ time?
The next ten years will see big data and machine learning principles permeate every aspect of technology and life. Ocado will predict when you’re likely to run out of milk, restaurants will have airline-like variable pricing based on real-time reservations analysis, and advertising will be a consistently relevant and valuable service, for both consumers and advertisers. I’m hoping to play some small part in a few of these transformations.
Who’s your tech hero?
He’s not the most unknown figure around, but I have an overwhelming sense of respect and admiration for Bill Gates. During his Microsoft years, he (amongst others) led the nerd revolution and showed the world that we could build some of history’s most successful businesses. In the years since Microsoft, he’s been a greater force for worldwide good, through his charitable efforts, than almost any person in history.
Who’s your tech villain?
I think the patent system, as abused by patent trolls and even major companies, is the biggest force currently working against innovation. With trivial patents being granted every day, it’s a veritable certainty that every tech company in existence is violating someone’s patents somewhere. As such, smaller companies are fearful of entering new markets or openly discussing their technology, more for fear of litigation than their IP being stolen.
What’s your favourite technology ever made? Which do you use most?
This may sound a bit trite, but my answer to both would be networking. Often overlooked, but completely pervasive, the real value of pretty much every modern use of technology has come down to networking. How useful are single computers in isolation?
What is your budget outlook going forward?
Struq continues to record fabulous growth and we’ve recently expanded into North and South America. We’re working hard to make sure this level of growth continues.
Apart from your own, which company do you admire most and why?
Although I’m not an active user of their service, I have an amazing amount of respect for Twitter. Like many of the new breed of massively scalable web services, they’ve had to build a lot of their infrastructure from the ground up. Unlike many of the others, they’ve open sourced a large amount of their work. Companies like ours, with growing scalability requirements, look to Twitter for inspiration… and code.
What’s the greatest challenge for an IT company/department today?
We’re at the crux of a paradigm shift in how IT companies are built and evolve. Five or ten years ago, a company like ours which had to build out massive business intelligence layers would provision racks and racks of SQL servers and convoluted middleware layers to achieve some approximation of performance. Now, technologies like Hadoop, Cassandra, MongoDB, Redis, Storm, and so forth allow similar objectives to be achieved, with radically simpler architectures (and lower costs). But we’re still in the early days for most of these technologies, and leveraging them requires highly skilled engineers who are intellectually fluid enough to adjust to a radically different discipline. IT companies investing in such tech will reap massive benefits, but they should certainly plan around growing pains.
To Cloud or not to Cloud?
For us, provisioning in the Cloud was a no-brainer. The last thing that a small tech company needs is to be distracted by things that don’t ultimately drive you toward your primary business objectives. With our hosting provider, SoftLayer, we’re able to quickly provision both virtual and physical instances and have them elastically integrated into our infrastructure within minutes. I don’t have to staff or manage technicians, keep supply cabinets of hard drives, or arrange real estate and cooling. Versus co-location options, we see significantly greater flexibility to dynamically scale our infrastructure and ensure our systems are always responsive and operating at peak efficiency.
What did you want to be when you were a child?
I was torn between wanting to be an astronaut and a mad geneticist. On the one hand, you had space travel, but on the other, two headed monkeys who could breathe fire.