Machine learning, idea factories, and data science are not typically present in standard accounting discourse. But these terms—along with others—are part of the BlackLine vernacular, as the company pushes to the outer edges of digital finance transformation.
Now working at BlackLine are Director of Machine Learning Sapna Nagaraj, data engineers Akshay Bala and Samarth Mothakapally, and data scientist Qixin Wang.
Akshay has a master’s degree in information technology with a specialization in data analytics. Samarth—or Sam—has a master’s in computer science. Qixin, who studied at the University of Southern California and the University of California at Irvine, has authored a number of academic papers on machine learning.
Each one came to BlackLine for the opportunity to work on challenging projects that will contribute to the future of the industry.
Their group is called Data Science R&D, and their mission is to find ways for artificial intelligence—machine learning (ML) in particular—to complement BlackLine’s products and services.
BlackLine Magazine: There’s plenty of talk in business about machine learning these days, but not so much in finance and accounting. One thing we do hear about in F&A is robotic process automation, though. What’s the difference between the two?
Sapna: RPA is a way to automate repetitive tasks that do not require a lot of thinking, but once demonstrated, can be repeated.
In data entry, for example, where you have to extract a few fields that are always in the same position in a file and load them into a database, there’s no thinking involved. It’s purely repetitive work.
But what if certain pieces of information are in the wrong location?
In that case, RPA won’t know that information is in the wrong position. It will blindly load the same piece of information in the same location without checking for errors. It doesn’t recognize that it should treat erroneous information any differently.
ML, on the other hand, has the intelligence to analyze data, identify where it needs to go, and act appropriately. The thinking part—encoding human knowledge—is why we bring in machine learning.
BlackLine Magazine: What are your functions within BlackLine?
Akshay: Sam and I deal with building out the fundamentals, or the building blocks. Once the building blocks are ready, then Qixin picks them up and builds the models for machine learning. Sapna architects the solution and guides the overall process to keep it on track.
We also have product people on the team who are constantly helping us understand the product—providing domain knowledge—and identifying opportunities to introduce machine learning.
BlackLine Magazine: How do you interact with other BlackLine teams?
Sam: We collaborate with our product managers, engineering teams, and analysts daily to brainstorm ideas and build robust software solutions.
We’ve created what we call an idea factory. It’s a shared virtual space where anybody in the company can go in and submit an idea based on what they think can be solved using automation or machine learning.
It’s open to everybody. It can be product people, accountants, implementation consultants, engineering, etc. Everyone has the opportunity to go in and write down their ideas in this shared, collaborative space. And, we ask them for a few basic pieces of information when they’re putting down their idea.
We run through the list, do a cost-benefit analysis to our clients, and pick the projects that solve some of the hardest problems for our users.
Qixin: We can even use ML to find out what features our customers are using and where we might help them get better value from our product features.
That way, we can develop smarter software to help people to do the things they don’t want to do, or they don’t have the time to do.
Sapna: This application of ML benefits BlackLine’s customers because it can help them use some features that they didn’t know about or hadn’t been using before.
BlackLine Magazine: Where do you foresee ML being used in BlackLine’s automation processes?
Sapna: It can be used in many ways. But right now, we’re focusing on bringing ML to Transaction Matching. Matching is a product that is very effective in speeding up the month-end close and reducing errors, and we want to get more customers to use it. We think ML will be a big help.
Read Machine Learning: A Brief Primer to discover why experts regard machine learning as a major branch of artificial intelligence.