NJIT Machine Learning Expertise Powers Tax Software for Vertex Inc.

Throughout the United States there are more than 11,000 tax jurisdictions collecting sales tax. Imagine it’s your job to keep up with all the changes to tax laws and help clients follow every new rule and file every return.

That’s the task Vertex, Inc. faces daily. Vertex develops tax software and services for indirect taxes such as sales tax, consumer use tax, payroll tax and value-added tax.

For years, many businesses manually calculated tax using spreadsheets.  Vertex has changed that by providing automated tax software. As a next step, the company recently partnered with the New Jersey Institute of Technology (NJIT) to use machine learning as a proof-of-concept to optimize tax calculation and further improve efficiencies.

“We’ve embarked on a program to embed machine learning capabilities into our products that serve the needs of businesses, as well as  the tax departments within companies,” said David Deputy, director of strategic development and emerging markets at Vertex. “We are looking to use technology to increase both the efficiency and the quality of tax determination and help our clients identify regulatory changes that may impact their organization.”

Vertex’s partnership with NJIT has two primary goals.

First, to reduce or eliminate the amount of redundant manual work for businesses. Deputy said that by using machine learning through its relationship with NJIT they have been able to use algorithms to pick up patterns for certain processes, which further automates and streamlines tax compliance. 

The second is to use machine learning to scan an enormous amount of information and find details tailored to benefit individual companies.

“Incorporating machine learning will be a tremendous advantage for Vertex, giving them the tools needed to sift through extremely large data sets quickly, allowing them to provide a unique and valuable service to their clients,” said Craig Gotsman, dean of NJIT’s Ying Wu College of Computing.

“By collaborating on the design of custom machine and deep learning solutions for Vertex, we have proven that our expertise is not limited to just theory but extends also to real world problems,” said Usman Roshan, associate professor of computer science and the lead professor on the project. Roshan and his team used open source tools that they also utilize in their academic courses to deliver a highly accurate tax categorization algorithm with the potential to scale to millions of data points.  

Deputy said Vertex is very happy with the progress of the partnership so far and the company is beginning to message clients to make them aware of the new options available. 

“The demand has been overwhelming,” said Deputy.

Deputy said NJIT’s location and the caliber of talent at the university made it the perfect partner for this proof-of-concept project.

NJIT offers graduate-level degree programs in data science, including a master of science in data science through the Ying Wu College of Computing, either at Newark or at its new location in Jersey City. Deeper research opportunities are available through NJIT’s new Institute for Data Science.