How machine learning and data science give Bloomberg a competitive advantage

CTO Shawn Edwards says strong AI capability helps the firm build data-led products for its customers.
Written by Mark Samuels, Contributor

In our meeting at the firm's European headquarters in London, Bloomberg CTO Shawn Edwards talks keenly about both the game-changing power of big data and his broader desire to build a data science capability that will help create innovative products for the firm's customers.

"I have an incredible job," says Edwards, who became Bloomberg CTO in January 2008. "I lead a relatively small group of researchers and our job is to help set the technology direction for the firm."

Edwards says his team gets to build proof of concepts -- working with Bloomberg's engineering department, product organisation and sales teams -- and generates new ideas, new architectures and new models. The team also works with a broad spectrum of external partners, including academic institutions, the open-source community, and IT vendors.

"When we think those ideas are good, we turn around and we socialise them," says Edwards. "And then, if we get agreement, we kind of switch hats and become the product owners. What's not to love about any of that?"

Edwards' CTO office is focused on finding innovation at the cutting edge of digital transformation. His team has the opportunity to search the globe for great data-led ideas and to put that research into practice in a business environment. In most cases, that effort is about finding creative solutions to business challenges.

"There is some exploratory work, where we think something seems interesting, and we try to figure out what to do with it," says Edwards. "But it's almost always the other way around: where we're driven by something that we want to accomplish, and then we focus on the best way to do it."

His team spends an increasing proportion of their day creating new products for customers. Edwards says the financial marketplace is changing -- there's more interest in a systematic approach to investing, which is based on models and rules-based algorithms. His team must think about how to serve this need through emerging technology.

"We're building a platform to help people analyse significantly more data than they've been able to do before," he says. "We're building programmatic interfaces into almost everything we do. Underlying all this, we're rethinking how we store and deliver our data. Being able to do a cross-domain query -- scan millions of bonds and join that with something about people, such as trading history -- is a game-changer in terms of the services you can create. It's pretty exciting."

SEE: 60 ways to get the most value from your big data initiatives (free PDF)

Edwards says machine-learning tools will be part of the suite of capabilities that his team is building for Bloomberg's customers. AI will help clients, like trading firms, to explore new data models and test hypotheses, such as potential changes in market prices. Customers will be able to take huge amounts of data, create new models and -- if it has value -- put this AI-based data model into production.

"We want to demystify AI, but really leverage it for it for what it's good at," says Edwards. "The tools allow you to build models and to build signals or extract value out of data that we couldn't five years ago. Now the tools are available, and what we're doing is making that all really easy to use for our customers."

The ongoing success of these programmes will be directly related to the quality of Bloomberg's in-house capability. Edwards recognises the ongoing challenge involved in attracting talent to run data-science projects. To help overcome this hurdle, Bloomberg runs a range of initiatives to find and source the next generation of data-science capability.

The CTO office works closely with leading academic institutions, including through a grant programme in data science. Edwards says Bloomberg pushes out annual calls for grants to students through a fellowship programme. The firm had about 60 applications for the programme this year.

"Through the fellowship programme, we sponsor a PhD student for a year -- we fund their tuition, we provide extra money for some of their research and then they come in and intern for us. That level of collaboration with the academic community helps us to recruit some of the best talent," says Edwards.

As well and grants and fellowships, members of the CTO office also present at top-tier institutions and business conferences, and that helps attract people. Edwards says his firm runs a significant on-campus recruiting effort. The recruitment team in HR, for example, partners with the engineering department and creates specialist campus teams.

"Recruiting comes in many, many forms," says Edwards. "Once we talk to people, and we bring them in and show them they can have a big influence on the industry and the company, then that hooks people into coming to Bloomberg."

SEE: How to implement AI and machine learning (ZDNet special report) | Download the report as a PDF (TechRepublic)

Edwards says the battle for data-science talent remains tough, yet it is one he and his firm are determined to win. New tactics continue to be developed, including training internal talent in the benefits of data science.

One of these projects exploits the skills of one of the key data scientists in Bloomberg's CTO office, David Rosenberg, who is also an associate professor in the Center for Data Science at New York University, where he teaches graduate-level machine learning.

"What we asked him to do was to create a career path -- let's create a training class to give a career path for engineers who have a strong math background and who want to start applying machine learning either in their department or in another department within Bloomberg," says Edwards.

The CTO office ran face-to-face classes in New York and London, and then put these lessons online. There are about 30 classes online now. Edwards says many of the lessons are focused on "pretty heavy stuff" -- any staff member who is keen to get involved would need a strong academic background. However, the benefits can already be seen.

"It's great when you have somebody internally who has the math background and has been spending time understanding the domain and knows about a subset of the financial markets. Putting all that knowledge together is brilliant," says Edwards.


How big data is helping to power Mercedes to F1 glory

Flash storage and advanced analytics allows Mercedes-AMG Petronas Motorsport to hone performance in simulations and out on the track.

Why the NHS is killing paper records to save lives

The NHS still relies on paper for many patient records. Getting rid of it could free up time, and money.

Cloud computing is the key to digital transformation at Guinness World Records

Dedication's what you need if you want to create a successful approach to technology-led change.

AI and machine learning: Building use cases and creating real-life benefits

One organisation, two very different uses of AI and machine learning.

NFL in London: How big data is helping to change the game when it comes to player performance

New technologies are giving NFL coaches, and football fans, better insights into how players are performing.

71% of UK businesses at risk due to data skills gap, CEOs say (TechRepublic)

Some 84% of CEOs age 25-34 see data as a major risk to organizations, while only half of CEOs over age 55 say the same.

Qualcomm bets $100M on new AI venture fund (CNET)

The company will look for startups focused on artificial intelligence for devices.

Editorial standards