Back to posts

Top of the MLOps

Greig Cowan is our Head of Data Science and Engineering for a team called Data Science and Innovation. He leads a team of Data Scientists and Data Engineers to research, investigate and build solutions that can help the bank and its customers. Whether that's building out our fraud detection system, helping fight financial crime or looking at how we improve the banks cyber security position. With the use of data, Greig’s team can better understand our customers and their needs - what they need and when they need it.

MLOps stands for Machine Learning Operations, but what is it?

Data science has been creating insight and value for the bank for a while, and Data Scientists have been using that data to build algorithms that can help customers, like in detecting fraud for example. A challenge has always been how to take the idea from the Data Scientists, and get it integrated into the way the bank operates - from a technology to process point of view. Doing that at scale, in a large organisation like this, can be challenging. MLOps is the response to that.

According to Greig, ‘MLOps is a way of looking at the work the Data Scientists do from more of a Software Engineering perspective. MLOps helps us build software products, taking them from development all the way through into pre-production, and then into maintinance over the lifetime of the product.’

The algorithms that MLOps taps into are living and breathing, in some respects of the words. ‘They adapt as new data comes in, so we must be vigilant and resilient to any changes. The systems must be robust. We’re able to spot risks, retrain the algorithm and then get it back up and running quickly.’

How does MLOps innovate for our customers?

MLOps is new for everyone across Financial Services, not just the bank. What we're doing with MLOps is innovating us internally - calculating what’s the best order to deliver it, what’s the best kind of processes to use, what people do we need to be able to report and therefore what skills do they need? We're constantly bringing new ways of thinking into the work we do.

But we’re also innovating in terms of the industry as well. Greig and his team recently presented the technology solution at an Amazon Web Services event and the feedback we got is that no one else is doing this on our scale.

Greig’s team doesn’t work with any one area of the bank, they have the freedom to roam and see how we can use data across all aspects of what we do. ‘We have strong links with a lot of different areas of the business, where we build a lot of trust between us. They understand what we can deliver, and we understand what they're looking for. That trust allows us to move quicker as a bank.’

In other cases, where it’s a relationship with a new area of the business, then it's all about education. What Greig’s team can do and how they work. To give context to the importance of the work, Greig’s team has almost quadrupled in size over the past three years. They had a project in 2021 which pulled together all our customer data into one location. We collect a lot of data, names, addresses, IP addresses, email addresses transactions, devices, this list goes on. But like LinkedIn, our customers are connected to each other in one way or another in a network, and Greig’s team can pass that onto the Fraud Investigations Team to help them with their investigations. It’s systems like this which help flag when you log onto a device from a new location, for example. To take that one step further, Greig and his team are applying MLOps on top of that to help us automatically spot this potential fraud.

Why should someone apply for a job with us?

We have around 90 million customers and processes a quarter of all payments in the UK. That means we have access to huge quantities of financial data on customers, payments, and transactions. To Greig, ‘there’s a massive data source so if you like working at scale, with a lot of information, working here is a real opportunity.

You can then use that data to really impact the lives of those 90 million customers, whether it's to keep them safe or trying to anticipate their needs and offer products that are more appropriate to them. You can really see that you have a direct connection with the customer.’

Work with people in the Data Science and Engineering space today.

Our roles in data