"Before Global Data Technical Operations existed, there wasn't a full-time role or a team in Data Management where you could be a technology project manager or a coder," explains Rob Keen, Team Leader of Technical Operations. Traditionally, Bloomberg's data analysts were responsible for sourcing and checking data relevant to their department - such as FX, equities or fixed income - and processing it on the Bloomberg terminal.
Over a couple of years, Rob and a few colleagues spent a lot of time talking about technical solutions that they could build to help improve the Data Management group as a whole. After realising that they could really make a difference to the department, Rob approached senior management with a proposal to make the team official.
Rob's pitch proposed quite a culture shift, as a lot of the key technical initiatives associated with the Bloomberg terminal were traditionally led out of Bloomberg's offices in New York or Princeton, New Jersey. However, senior management was open to Rob's recommendation. "One of the cool things about Bloomberg is how you are encouraged to be entrepreneurial," says Rob. "If you have a vision, you're encouraged to come up with ideas for how to get there."
Bloomberg's Global Data Technical Operations now work closely with data analysts to research and implement technical solutions to the issues the analysts face day to day. There are several different roles within the team, and many of them cross over.
There are those who project manage, who will usually be in direct communication with the data analysts; those who build prototype tools and solutions for the analysts to use; and those who do analysis to find any areas that need improving.
Joakim Nilsson is one of the eight specialists who make up Global Data Technical Operations and his role involves some analysis but also some coding as well. "I could be digging through big data sets to discover how our data analysts use their workflow tools and to understand where we can make the process more efficient. I could then take that research and build a trial desktop solution in VBA or use Python to model it. If it's successful, I could then be writing out a specification for our R&D department to build out a product solution for our analysts."
One project Joakim recently finished involved finding a way to increase analyst efficiency when normalising data in the Bloomberg terminal. What few people realise unless they use a Bloomberg terminal is that it also operates like a newswire, feeding a constant stream of stories from Bloomberg News and external news providers to its users. This feature is beneficial because it means analysts are well prepared to respond to any news stories (such as an IPO) and normalise any associated data on the Bloomberg terminal.
As analysts tend to have access to multiple sources reporting on the same event, they're often faced with a large amount of information on the one story. Many of these sources contain information they already know, yet because of the volume of information more than one analyst is needed to monitor the stories. In order to get around the problem, Joakim spoke to the various data teams to understand their needs, researched various solutions, and built a model in Python to address it. "We came up with an algorithm that can read the text of news stories and try to understand what they mean by analysing the sentence structures," he explains. "Now when two data sources that are very similar in content come through to us, they go to the same analysts for normalising."
"Before Global Data Technical Operations existed, we had no way of realising the common problems shared across the Data Management teams," explains Rob. "Now we have the ability to identify technical solutions across the department, as well as the opportunity to work in some cutting-edge areas."
Now he's transformed the technical capabilities of the Data Management group in Bloomberg's London offices, Rob has ambitions to expand the team further. "Asia is such an important part of the Bloomberg financial network. My ambition is to bring Global Data Technical Operations there as well."