Interview With First Tech Federal Credit Union: Scaling AI With Dataiku

Use Cases & Projects, Dataiku Product, Scaling AI Catie Grasso

It is a very exciting time in financial services and banking as the industry has become a commodity in many consumers’ minds and, particularly in the wake of disruption in 2020, people have never been more intent on digitizing their banking. Financial institutions are competing to remain relevant and differentiate themselves as a preferred and trusted choice for the consumer via scaling AI.

The team at First Tech Federal Credit Union (FTFCU), a premier credit union serving the world’s leading technology-oriented companies and their employees, has embarked on a mission to use advanced analytics to augment their understanding of their members (who they are, what their financial goals are, and so on) in order to deliver personalized and relevant experiences and become a truly data-driven institution.

Although First Tech is currently at the stage of integrating Dataiku in their platforms ecosystem, they look forward to transitioning their data initiatives to Dataiku once the development environment is ready. We sat down with Jay Franklin, VP Enterprise Data and Analytics, and Dr. Zack Protogeros, Senior Director of Advanced Analytics at First Tech Federal Credit Union, to learn about the company’s journey to Enterprise AI, the opportunities they are most excited about with regard to their overall data and analytics strategy, their evaluation process for finding an advanced analytics / data science platform and, ultimately, how they chose Dataiku.

Jay Franklin FTFCU

Jay Franklin leads First Tech Federal Credit Union’s Enterprise Data and Analytics Center of Excellence. He manages a team of advanced analytics/data science, business intelligence, and data management professionals located in corporate offices in both California and Oregon. He is well-seasoned for this role, given his background in business, IT, and consulting, driving large-scale data and analytics, business process, and technology initiatives and disciplines.

zp_pic_edited

Zack Protogeros, Ph.D., is the Senior Director of Advanced Analytics, Enterprise Data and Analytics Center of Excellence at First Tech Federal Credit Union. In his role, he focuses on optimizing processes through data science and analytics, and spearheads insight-driven interventions to delight First Tech Federal Credit Union’s members. 

Let's Get Started

Catie Grasso: How is your data team structured?

First Tech: We are the Center of Excellence for Enterprise Data and Analytics. Our charter includes delivery of high value enterprise analytics, as well as the establishment of data management practices (i.e., information governance, data quality and master data management, analytics enablement, and so on) to drive maturity of data and analytics across the organization. According to Jay Franklin, “Collaboration is critical culturally at First Tech. If we are going to become data driven, we need to promote a culture shift and work cross-functionally in order to enable teams to know how they can collaboratively produce more accurate and higher-value results.”

 

CG: How did you build your Enterprise Data and Analytics team?

FT: We established the Center of Excellence by identifying the critical roles needed to comprehensively articulate and execute a successful data and analytics strategy. We selected the team members by choosing individuals who show promise to excel and go the extra mile to advance the team and themselves. We focused on individuals with a strong background in data science and the motivation and aptitude to develop their soft skills to be able to effectively communicate their findings and recommendations to the wider FTFCU community and executives.

CG: How do you prioritize your data projects?

FT: Through a continuous due diligence process that starts by engaging the business stakeholders and follows through to identifying and precisely defining their business needs and pains. We make sure each initiative connects back to a business goal or pain point, makes sense, and can be quantified. We consider the effort (such as what resources are involved), feasibility, and expected ROI from any initiative and present these use cases to a cross-functional community of leaders to prioritize the efforts.

Zack Protogeros shared, “To us, good data science involves impactful engagements based on scientific rigor and lead to tangible results that contribute to delighting our members while rendering FTFCU a data-driven organization. Ultimately, good data science only occurs when the organization is willing and able to act upon insights obtained.”

CG: What is unique about doing data science in your particular industry?

FT: Being a credit union and not a bank, we are in a unique position to immediately and directly affect the lives of our members either by supporting them in achieving their financial goals at the appropriate stage in their lives, identifying which members may be a better fit to which services we offer, or by simply optimizing our internal processes. Such interventions become more pronounced and meaningful at times such as the COVID-19 pandemic. This has only increased consumers’ adoption of digital banking and expectations that their financial institution knows them sufficiently to provide personalized and relevant experiences and offerings.

CG: What did your evaluation process for a data science and machine learning platform look like?

FT: We extensively analyzed and evaluated a number of platforms using a standard set of features and functionalities offered by each one of the platforms such as connectivity to data repositories, automation options, collaboration options, etc. We offered our user base both a hands-on experience to use each one of the platforms, as well as demos. Finally, we normalized scores taking all entries into account.

We began speaking with Dataiku about a year ago as a critical component of our data and analytics strategy — we wanted an advanced analytics platform that could both accelerate our analytics capabilities and govern these critical assets. We envision 5-10 Dataiku users in the next couple of years, consisting mostly of data scientists and advanced analysts.

CG: What were your data efforts like prior to choosing Dataiku?

FT: A lot of data wrangling, lack of standards and best practices, lack of code versioning, auditing and visibility to code modules, lack of sharing and team collaboration, lack of documentation, and multiplicity of essentially the same data sources. Through a federated model, it is our hope that we will be able to create organization-wide standards (from credit and member analytics teams to marketing) on governance, data management, transparency, and so on.

When it comes to self-service analytics, we want people to be able to abandon cobbled-together dashboards and transition to doing things in a sustainable, scalable way. We expect the greatest contribution Dataiku will offer will be to generate assets out of artifacts, while providing the structure to protect the organization’s analytics assets.

Jay Franklin emphasized that “More often than not, the reason data science projects fail is not because the model is wrong. It’s because the business stakeholders didn’t know how to measure success or were unclear about the business role in executing the project or adopting the asset, once complete. We’re trying to teach not only how to identify something to work on but how it will help drive business objectives.”

CG: What are you most looking forward to with regard to combining Dataiku and Tableau? How about Dataiku and Snowflake?

FT: We are at the early stages of jointly leveraging these platforms. As far as the Dataiku and Tableau are concerned, one of the pronounced advantages has to do with integrating large amounts of forecast data with the excellent visualization capabilities of Tableau.

Leveraging the power and flexibility of Snowflake’s cloud-architected data warehouse with Dataiku will enable us to reference it as the authoritative source of truth for managing large datasets to support complex analytics use cases. We expect that both Tableau and Snowflake, together with Dataiku, will result in high-value outcomes that will benefit FTFCU and our members.

CG: Why were Tableau and Snowflake chosen for First Tech’s tool stack?

FT: Tableau was chosen because of their market leadership in data visualization, the strong R&D roadmap, and their overall standing in the Gartner Magic Quadrant for Analytics and Business Intelligence Platforms. Snowflake was chosen because of their true cloud architected and fully-hosted solution, strong customer references, and strategic vision.

CG: What will your first project with Dataiku be?

FT: We plan to deploy our auto loan delinquency prediction model on Dataiku as our first FTFCU production project. Risk score prediction allows the Special Assets Management (i.e. Collections) team to optimize their resource allocation by choosing the members most prone to default to be contacted first.

CG: What are you most excited about for the future of the Dataiku and First Tech relationship?

FT: Working with a Leader in the Gartner 2020 Magic Quadrant for Data Science and Machine-Learning Platforms will certainly allow our advanced analytics to flourish. The functionality offered and the maturity of Dataiku as a product will definitely support our efforts to render First Tech data driven, a result we believe will reflect on our members’ satisfaction. We are also looking forward to Dataiku as an organization being a strategic thought leader and partner as we continue in our advanced analytics endeavors. As we aim to reduce the number of software vendors we have and form deeper, more strategic relationships with them, we anticipate even greater value as Dataiku, Tableau, and Snowflake partner with each other to help First Tech be successful.

You May Also Like

Maximizing Text Generation Techniques

Read More

Looking Ahead: AI Hurdles IT Leaders Need to Overcome in 2025

Read More

No-Code ML and GenAI With Dataiku and Fabric

Read More

Unpacking 3 of the Biggest Controversies in AI Today

Read More