Dublin’s tech engineering prowess is making it an important European hub for AI and machine learning, says Zalando’s Sean Mullaney.
Sean Mullaney is the engineering vice-president of information at Zalando and the head of Zalando’s Fashion Insights Center in Dublin. He leads a team of more than 100 engineers and data scientists in the development of e-commerce data infrastructure and machine learning, which power a large part of Zalando’s digital experience.
Before joining Zalando, Mullaney served as the head of innovation for machine learning at Google Dublin, helping to establish the site as a centre of excellence for deep learning. Additionally, he spent three years working in Silicon Valley at Google X, Google’s research lab focusing on breakthrough technology, developing the drone delivery programme.
‘AI and machine learning are fundamentally improving the core e-commerce experience’
– SEAN MULLANEY
While at Google, Mullaney served as the head of big data and insights, leading the team that built next-generation sales systems that leverage large quantities of data, to enable sales consultants to effectively engage with advertisers. He also served as the head of sales operations for EMEA at Google Dublin, helping drive execution for a region with thousands of sales consultants and billions in revenue.
Prior to joining Google, Mullaney co-founded Masabi, a mobile train ticketing company that raised $20m in venture capital financing. He is also a founding partner of Inventive Capital, a venture capital boutique business that partnered with leading institutions such as Max Planck Institute.
What career decisions led to you leading information at Zalando in Dublin?
My superpower has always been data. While working at Google, I started building up a team focused on big data and building tooling for our sales teams to help them leverage big data. I worked in the Google X research lab for three years with the same team behind self-driving cars, Google Glass and other moonshot projects.
A couple of years ago, we had our third child and we really wanted to put roots down, and decided to come back to Dublin. I was really trying to find the next big opportunity, particularly to focus on machine learning, which has really become my passion. Zalando is so well known in Dublin for being one of the best places to practise data science. They were looking for someone to come and lead the office, particularly someone with a background in data and machine learning, but also in product, strategy, operations, retail – and those are all areas I worked in at Google for seven years.
How big is your team and what are the responsibilities of your role?
At the moment, we have 100 employees here in Dublin and I have another team in Berlin. I am responsible for the entire information architecture, information systems and machine learning that drive the Zalando e-commerce business.
To most people, Zalando is a fashion retailer but in Dublin, it is a data company. Obviously, Zalando being a retailer, we have huge investments in convenience and logistics; we have huge investments in buying fashion, our wholesale buying division; and then the digital experience is where I am involved.
I’m in charge of the Dublin office but really I’m in charge of information. I am responsible for all of the information platforms that drive the digital experience: our website, our mobile site and the app.
Why is data so crucial to Zalando’s DNA?
The two big data platforms are customers and products, and both of those platforms are built here in Dublin. We also have a really sizeable machine-learning team building foundational machine learning for fashion that enriches our customer and product databases with new attributes.
One of the most brilliant parts of the work that happens here is taking 300,000 different products that are online for sale at any one time and a huge stream coming online. The team here have built – using the dozen of high-quality images that we shoot for each product – a system that can actually learn about 125 different attributes about the product just from the images.
As these new products come on board, we are learning all of these things using computer vision. That means that every single product in our catalogue has a standardised understanding. For dresses, for example, it will tell you what colours are in the dresses, what the textures are, whether they are long, short, have arms, collars – all the things you would get from looking at an item, using computer vision.
How does machine learning help Zalando to sell more things?
Zalando is a marketplace for brands. We have thousands of different brands who all put their products into our platform. Each of those brands describes their product in different ways, different words, terminologies etc. When it all comes into our product platform, we need a way of standardising so that when people come to our site to search, explore, be inspired, we are able to give them the best results.
We have learned through first principles what these products are and it means that, in terms of the whole experience, we really are unlocking the power of our catalogue because we’ve created this insight using machine learning from first principles.
How do you manage the creation of data systems while growing at a significant velocity?
Zalando is still a young company; we are the biggest online pure-play fashion company in Europe but we only just turned 10 years old.
When a company grows that fast, there are still platforms that are put together where velocity of development is prioritised over scalability and, sometimes, the quality you need to operate at the scale that we are doing.
What the Dublin team has been doing has been rethinking entire implementation architecture about a company that is selling $5bn worth of product and is hoping to double in size in the next three years. To provide the information architecture for that company is completely different than the one where you are a start-up and growing really fast.
Data has been in Zalando’s DNA since the company was started. It was always an engineering and data-driven business from the start. Because it was online-first and online-only, we have really always had engineering and data. We were very early with machine-learning investments and I think we are ahead of the curve in terms of the maturity of our AI solutions.
From that perspective, we are in a position to grow in size and be able to offer digital experiences to customers that would be difficult for other companies to do.
What will the future of shopping look like, driven by AI?
10 years ago, people were talking about VR and how they are going to have bots and AI brought into mainstream commerce. I think we are still at a very early-adopter stage and I don’t think anyone really knows when those changes will happen, but they could happen next year, or five years or 10 years before we have truly intelligent systems and VR; when all of this more science-fiction-sounding technology becomes mainstream.
Today, AI and machine learning are fundamentally improving the core e-commerce experience. And I think we still have a long way to go to create a core e-commerce experience.
Our goal is to have unlimited fashion, so we want to be able to provide significant choice from all the brands and products you can possibly imagine, but to make the digital experience natural so that you are not overwhelmed by that choice. We can do that by using really deep machine learning and AI, and by having a very big set of data to learn from.
Hopefully, as Zalando improves and gets better, it is going to feel much more personal and things you are going to find on our site will be truly delightful and made for you, for what your style and taste is, but not overwhelming. We have a long way to go before we have that truly personal experience, where the core e-commerce experience knows you so well that you are really able to find those needle-in-a-haystack items that are totally suited towards you.
What role does Dublin play in Zalando’s journey from an engineering perspective? Is the talent available?
We have a phenomenal team here in Dublin, and that is recognised across the company. It is recognised in the core responsibilities the Dublin office has been entrusted with. We have really found a great team built up and developed here, and are delighted with the quality of engineers.
I think it is kind of a well-kept secret at the moment that there are fantastic engineers in Dublin. Quite a few companies in Dublin would have more than 100 engineers. When I lived in Silicon Valley, there were a huge number of expats living there who have phenomenal experience working in these big tech companies and I would love to see more of them coming back to Dublin because the opportunities exist. For Irish people in London, New York or California, the message is clear: there are opportunities in engineering, machine learning and AI now in Dublin.
Secondly, I think Dublin is becoming a tech centre for a lot of the EU talent, so not just the homegrown talent. Most of our office are Irish engineers who have been to Irish universities and represent local talent. But we really feel that we can recruit from anywhere in the EU as well, so there is a great talent pool by being in the EU as companies want to grow and scale. It is not just constrained to engineers coming out of universities here. You do have additional options.
Want stories like this and more direct to your inbox? Sign up for Tech Trends, Silicon Republic’s weekly digest of need-to-know tech news.