Natalia Lyarskaya, chief data officer, ZestMoney

Path to power

I started my career as a statistics analyst in 2006 in CRIF; at that time an Italian credit bureau and credit risk consultancy which was just about to open an office in Moscow. I spent the first months in Italy at the company’s HQ, getting trained on the new software (SAS) and learning the basics of scorecard development for the financial sector.

I then decided to pursue my studies and to do a PhD in Economics at the Sorbonne University in Paris. After completion in 2011, I moved to London and joined fintech start-up Wonga as its first data scientist building risk and fraud algorithms.

As the company was growing fast, I also built up my team and became lead data scientist, overseeing UK and international markets from the data and modelling perspectives. Later, in 2014, I created a new team, focusing on alternative data sources and emerging machine learning technologies for risk and fraud assessment.

In August 2015, I joined ZestMoney, an Indian fintech company with the mission to set-up its data science team in India, as well as to build its fully-automated decision engine using AI/ML technologies. In 2018, I was appointed chief data officer to take over strategic planning of all data and technology-related initiatives.

What is the proudest achievement of your career to date?

My greatest professional achievement was turning around the success of ZestMoney data science and credit risk teams. When I joined company, I was their second employee, and had to build the team from the scratch, as well as to develop policies and algorithms for real-time credit risk assessment and underwriting.

I’ve spent months in India looking for the best data scientists and engineers, while at the same time designing the risk engine and going through the in-depth research of available data sources on the market, running experiments to test user’s behaviour.

Within one year we had our very first automated decision engine built, which today allows to automatically approve/reject 95% of all applications within a few seconds. Currently, the team counts more than 20 members, taking care of ML algorithms, credit risk and fraud strategies.

Who is your role model or the person you look to for inspiration?

Having a hero/heroine helps you navigate through the difficult times. You look up to them and then think that the problems you thought were difficult are actually trivial in nature. If people can solve and deliver at a much larger scale, you can too.

There are a few data scientists who inspire me – Corinna Cortes (head of Google research, a competitive runner, and a mother of two), and Andrew Ng (a data scientist who also proved himself as a successful entrepreneur). ZestMoney’s CEO and co-founder, Lizzie Chapman, keeps inspiring me with her constant drive for learning, and ability to adapt to a fast changing environment.

Last, but not least, my mother is a continuous source of support for me, teaching with her own example that there is no limit to what I can achieve, however hard it could be.

Did 2019 turn out the way you expected? If not, in what ways was it different?

2019 turned out to be very successful. I fully restructured my team, splitting it into three sub-teams – data science, AI, and risk and fraud strategies – so we could more profoundly focus on each of the areas. There have been a lot of new learnings and changes on the Indian fintech market, starting from government KYC initiatives to significant shifts on the competitive landscape.

Despite of these turbulences, we’ve been able to achieve team goals by reducing losses by almost 40%, develop new algorithms and models, and migrate data to a new architecture. From personal development perspective, I’ve been part of several data-related conferences, including Women in Data (UK) and CDAO (Berlin), allowing me to share my experience and increase my professional network.

What do you expect 2020 to be like for the data and analytics industry?

I believe 2020 will be a year when big data and AI/ML technologies will become a crucial part of many companies, across industries, to stay competitive and be able to achieve their goals. Many companies will move beyond just POC stage when applying big data technologies, and fully integrate them into day-to-day processes and products.

It will no longer be an isolated effort; rather companies will start looking at data and technologies in a holistic way with all the consequences of such a purpose-driven and integrated approach. More and more services and products will appear in the data and analytics segment, though markets become more and more matured, leaving behind less accurate and less competitive products.

Data and technology are changing business, the economy and society – what do you see as the biggest opportunity emerging from this?

Data and technology are no longer considered as cost centres, but profit drivers and revenue generators; this in turn changes the established business models, forcing companies to better respond to market demands, customer requirements and investor expectations.

The biggest opportunity fostered by data and technology is creative innovation – innovation in products, processes, and the way companies respond to changes in external environment. Just solving the end user’s problems is not enough. One needs to take into account how the problems are solved – are solutions scalable, is end-to-end process well thought out and digitalised, can it be easily replicated by the peers?

What is the biggest tech challenge you face in ensuring data is at the heart of your digital transformation strategy?

Data and analytics are pointless if they can’t influence the decisions. The biggest challenge is to ensure data and technology-driven culture is spread and fully understood across all levels and among all the stakeholders and is not merely concentrated within data science and engineering teams.

Every business decision should be supported by data, analysis, experiment results, and not by gut feeling, legacy considerations, or prior experience. This, in turn, is closely related to transparency that data and AI can provide to human experts, ie. explainable AI. “Black box” logic is no longer sufficient, neither satisfactory from business and regulatory viewpoints to be used in a BAU manner. The results of the data analysis or AI solution need to be fully understood, replicable and interpretable by the decision maker.