Set-up in 2003 in India as a Global Capability Center (GCC) of Fidelity Investments. Fidelity Investments India has offices in Bengaluru and Chennai. Fidelity Investments, India functions as a microcosm of all businesses in the U.S. market, providing strong support to the global business as well as its customers. Fidelity Investments is an American multinational financial services corporation based in Boston, Massachusetts. They started out in 1946 as a small mutual fund company and now, they are one of the largest financial services firms in the world. Fidelity Investment's core business areas include Asset Management, Personal Investing, Workplace Solutions, Technology Services, Enterprise Delivery & Solutions, Business Analytics & Research, and Support Functions.
Analytics Insight has engaged in an exclusive interview with Mridul Mishra, Vice President of Emerging AI group, ARD, and Fidelity Investments India.
Fidelity Investments is a privately held financial services company founded in 1946 and our goals have been "to strengthen and secure our customers' financial wellbeing, and to bring unparalleled value to the people and companies we have the privilege to serve". Fidelity is one of the largest asset managers in the world, with $4.2 trillion in assets under management as of Q3, 2021, and a combined total customer asset value number of US$11.1 trillion.
We help over 40 million people to feel more confident in managing their financial goals, manage employee benefit programs for over 22,000 businesses, and support more than 13,500 financial institutions with innovative investment and technology solutions to grow their businesses.
Our diverse businesses and independence give us insight into the entire market and the stability needed to think and act for the long term as we deliver value to customers.
Fidelity Investments India is a microcosm of Fidelity Investments, with all of the business lines represented here. We are a strategic team that delivers on enterprise objectives, managing cutting-edge work for all businesses in collaboration with colleagues across the globe. In India, we are structured across three key functional areas – Technology, Analytics, Research & Data (ARD), and Operations.
Technology has been always been a major focus area for us at Fidelity. As our CEO Abby Johnson once said, "Fidelity was doing fintech before fintech was cool." Over the years, we have invested significantly in new and emerging technologies like blockchain, artificial intelligence, voice recognition, and virtual reality, to name a few.
The ARD group in India is a CoE focused on servicing fidelity business units with advanced analytics and data science capabilities. The group has focused on multiple analytical competencies such as advanced analytics, visualization, time series forecasting, natural language processing (NLP), computer vision, neural networks, and reinforcement learning, to name a few.
ARD has also focused on building domain capabilities in the areas of investment decisions, operational analytics, people analytics, cybersecurity, and fraud prevention. The group has built significant experience and capabilities in primary and secondary research as well.
Big data, analytics, and artificial intelligence (AI) have moved from the stage of "Peak of Inflated Expectations" to "Slope of Enlightenment", according to the Gartner Hype Cycle. We are currently witnessing a significant progression from a proof of concept (PoC) stage to the large-scale deployment of big data/ analytics /AI-enabled solutions. The operationalization of these solutions has begun delivering transformative business value.
At the same time, we have observed that these solutions are not "silver bullets" and should be matched with appropriate use cases. We often witness organizations pursuing trending technologies as the panacea for all problems. However, from our observations, we have learned that an optimum solution doesn't always lie in the latest technology or algorithm. If you have an in-depth understanding of the business problem, using an existing technology might sometimes deliver optimal results.
As the focus shifts to operationalization, we are witnessing a significant emphasis and investment in a platform approach to data, compute, deployment, governance, ethics issues, and a need for a well-defined, consistent process. We can find similarities in how DevOps emerged as a philosophy in software development and how a well-defined set of tools and practices enabled organizations to deliver at a higher velocity. Similarly, the emergence of fields such as Xops (AIOps, DataOps, ModelOps, etc), will help in operationalization at an accelerated pace.
We are witnessing three significant trends driving business value using big data analytics/ AI/ machine learning (ML).
Although supervised learning is still the workhorse in enterprise implementations, of late unsupervised learning implementations have begun making headway. This trend is especially prominent in natural language processing (NLP). Developments in unsupervised learning, to an extent, are solving the problems associated with a lack of labeled data. Pre-trained models and transformer architecture like BERT in NLP are also significantly reducing training time and expediting value delivery.
We have noticed significant progress from cloud vendors on providing AI/ ML capabilities as a service, especially in NLP, (Amazon Web Services [AWS] Comprehend, Azure Cognitive Service for Language), vision (Google Cloud Vision [GCP]), and forecasting (Amazon Forecast). In the past, most of these offerings were custom-built on the premise using limited infrastructure and data by scarce data scientists. Today, however, these offerings are available as managed services and can be used to create more valuable solutions, thereby significantly reducing the value delivery time. The availability of almost infinite, on-demand, cost-effective compute and storage has been foundational for these use cases and critical for faster value delivery.
The adoption of cloud and data-sharing platforms has reduced the value delivery cycle. These platforms have reduced the cost of storing large amounts of data, which is a foundational element for building analytics and AI solutions. Today, we require significantly less time to understand feed structure, load data and create useable solutions. The optimization of the value delivery cycle is critical for rapid experimentation as investments in a PoC can be operationally more efficient.
Fidelity is a pioneer of mutual fund investing, workplace retirement planning, discount brokerage services, digitally based financial services, and more. Our process of rapid and iterative product innovation, based on the agile method of software development and project management, helps the company to anticipate the evolving customer preferences and to build innovative products and services. Our focus on delivering a superior customer experience across various life stages helps produce growth in the number of customers and accounts, which then allows us to drive scale and efficiency and reinvest in product innovation.
As a result, we are developing next-generation digital platforms and capabilities to better serve our customers, such as, for example:
In investment management, AI and analytics have helped businesses to achieve a breakthrough transformation in identifying the next big investment opportunity, performing efficient trade executions and delivering superior customer experiences by providing personalized recommendations. Traditional customer-facing processes such as onboarding, change management, and withdrawal have gained significantly from the integration of AI and analytics into the workflow.
In the risk and compliance area also, digital transformation is strengthening several areas, ranging from insider trading identification and fraud prevention to communication surveillance.
Machine learning is rewriting the rules of the IT and financial services industry. As the cost of storing and processing data is diminishing and the availability of advanced algorithms is increasing, the current generation of solutions is capable of identifying even more complex relationships between varied datasets and making better predictions. The impact of computer vision and NLP/ natural language understanding (NLU) techniques is growing significantly in most industries, including financial services.
Most of the large financial services organizations today run on legacy systems. There is no doubt that these systems are getting modernized. However, since upgrading legacy systems happens in phases, it may be challenging at times to keep up with the rapidly changing landscape of AI and integrate the new technologies into current systems.
With rapidly advancing technological capabilities, the role of C-suite leaders is changing at an exponential pace. Traditionally, large organizations would focus their technological efforts on keeping their systems robust. However, today, the efforts are shifting towards adapting capabilities like AI and big data analytics.
It is especially crucial for financial institutions to keep up with the pace of technological growth while ensuring data security and accordance with regulatory compliances. To stay ahead of the curve, CIOs will need to be aware of the impact of a potential new technological advancement and identify opportunities to integrate these technologies into internal systems organically or through mergers and acquisitions.
The scenarios mentioned above are only a few of the challenges and opportunities CIOs will face. To summarize, the CIOs as well as the other C-suite leaders will need to play a holistic and strategic role in planning growth strategy, talent strategy, and organizational design, rather than just being the gatekeeper of technology architecture for an organization.
While earlier AI/ML algorithms were developed and delivered in a bespoke manner, the growing need for scale has mandated the need for focusing more on automation, platforms, and consistent processes like ML Ops, and DataOps. Traditionally, most of the monitoring, retraining, and subsequent deployment used to be manual processes that required human effort and had a potential risk of omission. The advancement of technologies has now enabled organizations to use MLOps capabilities like model management and drift monitoring to create an automated model management pipeline. Data platforms like enterprise data lakes have also made scaling and value delivery significantly faster and more efficient.
Cloud service providers like Azure, GCP, and AWS are currently providing some of these AI/ ML capabilities. However, the cloud services by these providers solve only a part of the puzzle. The rest of the puzzle involves an organization's capability to get data readily available for the model – in a data lake. Fidelity is strategically invested in ensuring a common, enterprise-wide data lake that can scale AI/ ML efforts and deliver value quickly.
AI/ ML/ analytics faces multiple challenges in growth and adoption. These can be broadly categorized into technology, people, and process challenges.
In terms of technological challenges, especially in financial services companies, explainability—the ability to explain why a certain outcome makes sense to a layman—becomes a foundational expectation. Explainability is not just a regulatory ask but a fiduciary responsibility. Although there has been a significant improvement in performance measures due to advancement in deep learning algorithms, explainability still remains a challenge in promoting large-scale adoption. Another challenge is to ensure fairness and prevent biases to be able to extrapolate models. At Fidelity, we have defined an internal process to ensure model fairness and adherence to ethical standards.
The second major challenge is the lack of availability of historical, good-quality data. Fidelity has strategically invested in data and continues to focus on it. Designing processes related to data availability and management is also a crucial part of the equation. One of the key challenges involves the process of aligning analytics/ AI PoCs with enterprise technology strategies for scaling. Historically, we have seen rough edges around the integration and migration of these PoCs and have now fine-tuned a process to address it.
The third key challenge is around people. Identifying the right problem to be solved is the key. Stakeholders need to be educated and sensitized that finding the right solution for a problem is critical to success. Chasing the latest fad, just for the sake of it, can be inefficient and costly.
