Inside ProDev: Data Science Takes Centre Stage
Dr Douglas ‘Data’ McIlwraith is Unruly’s resident data scientist. If you want to know more on machine learning, experimental methodology or ad optimisation, Dr Doug is your man. Aside from his work with Unruly’s Product Development Team, he’s also helped bring the world’s most prestigious AI and data conference to London. We sat down with him to find out more…
MS: Hi Doug. Can you tell us who you are and what you do here at Unruly?
DM: Certainly! I’m Dr Douglas McIlwraith, resident data science consultant for Unruly. I support all aspects of data science across the company – from working with our product development team on machine learning pipelines for advert optimisation, to helping our UnrulyEQ team develop the best experimental methodology. There’s so much to do, and we are only just scratching the surface of what’s possible.
MS: Talk us through a typical day.
DM: My day-to-day role is pretty varied, but it’s always focussed on pushing ourselves to make the best use of our data, and to create the best data products in market. I typically work with all areas of the business on this, from members of the executive team to product managers, software engineers, data engineers and other data scientists. It’s all about highlighting the possibilities of best practice in data science and allowing us to make the best product decisions so that, ultimately, we can go to market with the best data products in video advertising on the planet!
MS: Talk us through your connection with KDD2018. What is it?
DM: Sure! So KDD stands for Knowledge Discovery and Data Mining. It’s widely recognised as one of the leading global meetings in the areas of machine learning, artificial intelligence and data science. It attracts up to 3,000 attendees with some of the most well recognised names in these fields.
Every year it moves to a new location, and cities are invited to bid to host the event. Two years ago, and with the support of the Unruly founders, I got together with Duncan Ray from Hackney council to place a bid for the event. I was blown away when I heard that we’d beaten some unbelievable competition to bring the event to London for the first time. I have a chair on the organising committee and I’ve worked closely with the other members to make this event the biggest and best data science meeting in it’s 24 year history!
MS: Sounds impressive. So what makes KDD2018 such a big deal, and what kind of issues will be discussed?
DM: KDD is so important because it is a truly interdisciplinary and applied research meeting. It’s not just theoretical insights, you’ll also get to see these ideas applied to massive datasets as well. Google, Facebook and Amazon are all regular contributors, and have a huge presence in our exhibitors hall. It’s really the best place to see the latest advances being worked on by the tech giants.
MS: Why should ad and media people pay attention to KDD?
There’s always a significant focus on advertising and media. I’m particularly looking forward to an invited talk by Dr Suju Rajan, Head of Research at Criteo who is going to discuss the technical challenges of computational advertising at scale. Criteo have been been making groundbreaking advances over the years, so just to hear their take on where we are now will be extremely valuable.
Suju is also running an amazing workshop called AdKDD: it’s a whole day dedicated to data science for computational advertising. There are also some really interesting advances in the areas of header bidding and optimal allocation of inventory from an SSP perspective in the main conference track, not to mention the Data and Journalism workshop that I’m organising.
MS: What’s the story with the workshop?
Dan Gilbert, who is Director of Data at News UK, and I joined forces to propose a whole day workshop on data in the media, as it seemed appropriate given the large number of media organisations in London. We ended up broadening the scope and joining up with other experts in the field to cover journalism as well. Now we’re working with organisers from Bloomberg, Darmstadt University, Zeit Online, IBM and the University of Illinois. We’ve got some great keynotes lined up. I can’t wait to hear what Maria Mestre from Factmata has to say about their work on fake news detection. It’s exceptionally relevant right now as advertisers are keen to ensure that they are not associated with poor quality content.
MS: What are you most looking forward to about KDD?
DM: I’m really excited about Deep Learning Day, which is a new initiative. They’ve got Andrej Karpathy, Director of AI Tesla giving a talk, so if you want to learn about the latest advances in self driving cars, this could be your chance. There are many, many other exceptional speakers, but Nobel Laureate Alvin E. Roth is definitely worth calling out. He won the Nobel Prize in 2012 for “for the theory of stable allocations and the practice of market design”. This makes him spectacularly well qualified to deliver his keynote ‘Market Design and Computerized Marketplaces’. Given the growth and ubiquity of new marketplaces that are facilitated by smartphones, think AirBnB, Uber, Deliveroo, Didi etc, this is a really timely talk that I’m sure will be extremely illuminating.
MS: Finally, data science rules because….
DM: Data science rules because it has so much potential to be disruptive and transformative. We are seeing huge, positive advances in so many businesses and business verticals that have been enabled by access to the vast quantities of data we generate. It’s such an exciting time to be involved in the field. I’m really looking forward to pushing ahead at Unruly and seeing how we can use data science to provide the best experience for our clients and users.
