Outside of the technology sector, banking and insurance are perhaps the biggest employers of data scientists. Given that financial services have always been dependent on data and models – for example, for loan approvals or insurance underwriting – this should not be surprising. As a data scientist, though, how do you decide if it is for you? What real-world problems will you be solving? What challenges can you expect? What does it take to succeed?
Five good reasons to join…
One of the best things about a data scientist’s job in financial services is the sheer richness of use cases and the real-world impact a data scientist can have. Of course, there are all the usual applications common to any customer-facing business, such as personalised experiences, targeted cross-sell offers, or proactive strategies to prevent customer churn. But banks, insurers, and their fintech challengers use data and analytics in many other interesting and impactful ways.
- Use of alternative data, extracted from sources such as satellite imagery, telecom providers, IoT networks, or social media, to improve the pricing of insurance risk, extend lending to traditionally excluded customer segments, or generate investment ideas
- Network analysis to create detailed pictures of client ecosystems – customers, suppliers, staff, and owners – that help detect and investigate instances of terrorist financing, money laundering, or human trafficking
- Natural language processing algorithms to detect potential insider trading in an investment bank
- Image recognition algorithms to automatically process most auto-insurance claims
- Calculation of the environmental footprint of a proposed financial investment, such as a new industrial facility or even an individual mortgage
A second attraction for many data scientists will be the breadth and depth of data sets that can be made available to generate meaningful insights. Banks and insurers often have access to extensive data, such as demographics, transactions, and relationships, at both a macro-level and at the level of individual customers. Despite some restrictions on their use, the availability of high-quality data sets like these, often going back years, can be a data scientist’s dream when it comes to building predictive models.
The size of financial services firms’ spend on data and technology and the relative maturity of their data ecosystems can also make them attractive for data scientists. For example, most banks spend more than 10% of their annual revenues on technology. Spending on data and analytics is an increasingly important component of that and can easily reach or exceed hundreds of millions per year for many of the larger players – a figure unmatched by all but the biggest in the technology sector. As a result of years of spending on data, many also have relatively mature data teams. As a result, data scientists might find well-established support systems in place and are not expected to manage everything from data pipelines to data governance all on their own.
Lastly, in most geographies, banks, insurers, and fintechs are often some of the best paymasters for data scientists. While attractive in its own right, it is also a useful indicator of the value that is being placed on data science in these firms and the implications it can have for longer-term career tracks. In at least one major global bank, the Chief Data and Analytics Officer now reports directly to the Group CEO.
… And a few challenges, too
There is, of course, a catch. Everything that makes it interesting to work as a data scientist in banks and insurers, particularly the bigger ones, can also make it unwieldy and frustrating at times. Some data scientists will view these purely as challenges; others might also view them as opportunities to develop themselves and have a greater impact.
Given the high-stakes use of data and analytics in the industry, there is a high bar of trust to prove that data and models are good enough for real-life use. For example, if a data scientist is building a predictive model that could be used to deny somebody a loan or insurance cover, or mark someone as a potential money launderer, then they should probably expect a huge amount of scrutiny.
Similarly, given that customers often trust banks and insurers with the most intimate aspects of their life – for example, their income or their medical histories, data scientists can find elaborate controls around data availability and usability. Concerns around data privacy, sovereignty, ethics, and security exist in every industry, but very few other industries spend as much time and effort on managing them.
Extensive spending on data and related technology, and well-resourced teams of data engineers, analysts, and risk specialists can provide fertile ground for data scientists to thrive. However, the same factors can also lead to loss of agility on a day-to-day basis. In many cases, these might translate into restrictive technology choices for data scientists or multi-step processes with elaborate controls and hand-offs before their work can actually see the light of day in production. One particular area that surprises new joiners to banking is the need to get all material models formally validated by an independent team – a step that can add weeks or even months to the normal model lifecycle.
Underpinning all of the above challenges is the fact that financial services are one of the most regulated industries worldwide. In response, most banks and insurers have built a DNA of risk and compliance, particularly after the 2008 financial crisis. In many geographies, senior managers at banks and insurers bear personal accountability for the actions of their employers, so anything that holds the potential of breaching customer trust or regulatory requirements is treated with particular caution. The use of data and algorithms ticks all the boxes. Not surprisingly, financial regulators have been among the first to come out with guidelines around responsible use of data and AI – for example, in Singapore, Hong Kong, the European Union, the United Kingdom, and the United States.
So is it for you?
Clearly, not every data scientist will like it at a bank, insurer, or even a regulated fintech. However, you should actively consider a career in the industry if:
- You are excited about the opportunity to use your skills in such a broad range of real-world applications and make a meaningful difference through initiatives such as greater financial inclusion or better targeting of financial resources to support the global climate agenda.
- You believe that your job is as much about educating others about the ‘magic’ that is data science, and winning the trust of non-data scientists, as it is about building great models. You might be the expert, but there are many others in your firm who are not and still must put their names behind your work.
- You accept that a certain level of standardisation and discipline in “ways of working” is a necessary price to pay to have a large-scale impact from your work.
- You view restrictions around data use, such as those related to privacy, ethics, and sovereignty, as not just challenges or irritants but as opportunities to do the right thing and to develop yourself personally. For example, banks and insurers have been some of the earliest adopters of privacy-enhancing technologies. Most are also at the forefront of algorithmic transparency and fairness initiatives. Working at one of these firms can provide you, as a data scientist, hands-on opportunities to build models that are high-performing and trustworthy.
Bios: Shameek Kundu is a leading expert in AI from both a tech and business strategy perspective and has spent most of his career driving responsible adoption of data analytics/AI in the financial services industry. He is Chief Strategy Officer and Head of Financial Services at TruEra. He sits on the Bank of England’s AI Public-Private Forum and the OECD Global Partnership on AI, and was part of the Monetary Authority of Singapore’s Steering Committee on Fairness, Ethics, Accountability and Transparency in AI. Most recently, Shameek was Group Chief Data Officer at Standard Chartered Bank, where he helped the bank explore and adopt AI in multiple areas (e.g., credit, financial crime compliance, customer analytics, surveillance).
Divya Gopithan is a Research Engineer at TruEra, a company focused on making AI trustworthy and transparent. Prior to joining, Divya completed her undergraduate and Master’s degrees at MIT, where her research focused on building machine learning algorithms for healthcare domains. Divya is a major contributor regarding Trustworthy AI on Towards Data Science, focusing on topics of fairness and addressing bias in machine learning models.
Arridhana Ciptadi is a member of the engineering team at TruEra. He was previously part of the founding team at Blue Hexagon, where he was technical lead for all machine learning efforts for the company. Prior to that, he was a Machine Learning Scientist at Amazon Lab126 where he developed machine learning and computer vision technology for various Amazon products. Ciptadi holds a Ph.D. in Computer Science from the Georgia Institute of Technology.