Lucian Vlad Lita is Director of Data Engineering for Intuit, the well-known business and financial management software company based in Mountain View, Calif. He leads the big data platform and large-scale, real-time data services group in the US and EU.
Previously, Lucian founded Level Up Analytics, a big data and data science firm, before it was acquired by Intuit in 2013. And at BlueKai, he led the engineering and analytics team focused on big data, real-time audience management and analytics.
icrunchdata speaks with business leaders about their careers and latest focuses. Today, Lucian discusses bringing together scalable services with machine learning, building personalized experiences and hiring standout data scientists.
It was a seamless transition in that we were building intelligent data products and platforms then, and we’re doing the same now. The core difference is that now we can scale our work across an entire product ecosystem, with an end-to-end data strategy and deliver a stronger benefit to the end user.
At Level Up Analytics, we interacted with over 100 companies, ranging from startups to large companies, and from social networking to hedge funds to mobile and music streaming companies. We invested significant time and effort to understand our clients’ products and what they excelled at. It turned out that what they asked for and what they actually needed were often two very different things.
Over time, we gained a lot of insight into how companies are scaling with data as well as how they are using data to feed it into products and guide the user experience. We gave our clients practical advice around data as well as how to use data science and data engineering to build and improve their platforms and products. However, as a third party, it was difficult to create a fast, continuous end-to-end feedback loop, going from data insights, to user experience, to user actions, and then all the way back to better insights.
When Level Up Analytics was acquired by Intuit, we had the opportunity to mold over 60 products with the power of data, and ultimately take the user experience to a different level. Suddenly, we had the ability to access and leverage data across the entire financial landscape, so that we weren’t building one-off solutions for individual products, but rather a scalable data platform and services relevant across Intuit’s entire financial ecosystem. That’s 37 million customers’ lives – across QuickBooks, TurboTax and Mint – that we could impact.
Imagine a world where we get your data to work for you. When you start to prepare your taxes in TurboTax, we can remind you that you forgot to add the interest form for your Fidelity account, since we know that from your Mint.com day-to-day use. Since you’ve entrusted us with this integration in your personal finance management, we can retrieve the right forms automatically for you, so you don’t have to. Or better yet, we can retrieve forms proactively for you, automatically process them with OCR, extract the relevant information, and fill out your tax forms ahead of time. This is just one example where we bring together scalable, fast services with machine learning to make your data work better for you.
A few years ago, the Intuit Data Engineering and Analytics (IDEA) team was founded to support Intuit on its transformation into a cutting-edge technology company with data as a strategic pillar. Intuit is the steward of data from 37 million customers, and we’re rapidly growing in the US and globally. The IDEA team’s goal is to harness the power of this data at scale by creating a data platform and real-time data services to transform our products – and ultimately our customers’ financial lives. End-to-end scalable software architectures and machine learning play a vital role in achieving deeply personalized user experiences, changing the way we collect and process data, and how we then determine what useful, actionable nuggets of information we can deliver quickly to the user.
Basics first. We set out to build a big data platform to derive insights from all products. This meant working with all of Intuit's product groups and bringing in all of their data to fuel analytics and to impact products.
Next, we built real-time distributed systems that are powered by data. In most cases, these have to be under 60 milliseconds fast because they directly power the in-product user interaction. Examples of these systems include A/B testing as a service, clickstream capture, user entered data stream processing, business search, and graph – Intuit sees almost a quarter of the US GDP flow through its products.
The Intuit Analytics Cloud (IAC) is a petabyte-scale big data platform we’ve developed. It has a robust, scalable infrastructure that absorbs, processes and analyzes data from Intuit products, customer interactions and third-party sources. The platform targets an information-rich profile of our customers across all product experiences and touch-points to enable deep analysis and modeling, fueling personalized, relevant product experiences for our customers.
What sets the IAC apart from other big data platforms is the sensitive nature of the data and our support for data governance, compliance and security to make sure we protect our customers’ data.
I’ll share two examples of how we’ve delivered personal experiences for customers, thanks in large part to our big data platform, the IAC.
We learned from our data that 60 percent of QuickBooks customers (global small businesses) have been denied loans. This is often due to the lack of complete, accurate data that financial institutions need to approve loans. QuickBooks Financing uses data that is already part of users’ accounting system and provides lenders a view of small business risk. It’s more complete than the traditional personal FICO score, matching qualified businesses to the right lender. To solve this problem, we’ve developed algorithms that leverage the IAC to extract value from QuickBooks data previously locked in the product. As a result, the percentage of small businesses approved for loans has drastically increased. Targeted campaigns in QuickBooks Financing have seen a 70 percent acceptance rate and small businesses have secured more than $200 million in credit. This is incredible when you think of the impact to small businesses trying to survive and grow.
TurboTax is another product that uses the power of the IAC to deliver a great customer experience in-product when they need it most: during the tax filing process. Using the IAC, we combine TurboTax customer care data from different channels – in-product, social, community, chat and phone – and we zoom in on the relevant insights and data that helps us improve the product. Armed with the knowledge of which customers may be getting confused and which screens they are using repeatedly before contacting support (made possible by algorithms known as path analysis), TurboTax teams can make screen-level improvements with greater confidence and speed. These product improvements empowered 415,000 additional customers to complete their taxes entirely on their own, without the need to call customer support.
Machine learning and predictive modeling algorithms are important, of course. However, there are two other ingredients that are far more important early on:
Our goal is to continue personalizing our customers’ experience across all of our products. In QuickBooks, that means a personalized landing page that’s optimized for their workflow. For TurboTax it means providing meaningful evidence that their taxes are filed correctly. Building one model, or one product feature, at a time will not get us there. However, a robust platform that can support massive data movement and processing quickly, and the ability to train, deploy, and manage predictive models at scale will.
For the IDEA team, ultimately, it's about the customer. It's about making their lives better. And the smarter we are with the data they are entrusting us with, the more we can help them save money, get a bigger tax refund, run their small businesses, grow, and thrive. Personalization is a big step towards achieving these goals.
Traditional approaches to data science and analytics gravitate toward a centralized model, where all requests are funneled to a single team. This was a good idea in the early days of data science, but we have entered a different age, one where the only way to scale is to democratize data, data processing and analytics. In other words, build a platform, provide strong security and data governance guardrails, and then open up the magic to all product groups in the company. That’s our vision with the IAC – to put data processing and analysis tools directly in the hands of our analysts, engineers and scientists.
My hiring philosophy relies on four skills that I find crucial in building a great team:
I am not looking for candidates to be experts in particular technologies or methodologies. The industry moves too quickly for us to think this way. Hire great people who can learn and adapt. Make your own experts!
So this combination (in that order actually) is what makes, in my mind, somebody I would love to work with and learn from to make a difference together.
I can’t think of any company that does not or should not use data. So at this time, any company I envision would consider data as part of its core capabilities and see it as a strategic advantage.
I’d love to do something in the education space. There are already several very creative and driven companies and non-profits that are challenging the status quo, but the field is still ripe for disruption. There's still room to push and radically change how education is imagined and delivered. People learn in different ways. For example, some people can learn amazingly well from textbooks. Others are more visual and need to make additional connections with the topics at hand.
That's it for now. Thanks, Lucian, we've enjoyed getting to know more about you and your data insights.