Student Startup Uses AI to Modernize Legacy Code

For many companies, decades-old software code can be more of a liability than an asset. As technology rapidly advances, maintaining these outdated systems—built using programming languages that are no longer taught and supported—has become an expensive and time-consuming challenge. In today’s fast-paced environment, businesses struggle to keep up with competitors due to the constraints of legacy systems that were once at the forefront of innovation but are now holding them back. 

Axal, a startup formed by University of Maryland students, seeks to address these challenges. Led by UMD computer science majors Samai Patel (B.S. ’25, computer science), Nand Vinchhi (B.S. ’27, computer science) and MIT student Veer Gadodia (B.S. ’24, computer science), the company developed a platform that uses artificial intelligence to streamline the process of understanding and modernizing outdated systems.

Axal’s recent entry into the 2024-2025 cohort of the Mokhtarzada Hatchery Program at UMD’s Department of Computer Science allows the team to refine its approach and access mentorship and resources that once were inaccessible. 

Inspiration

The idea for Axal emerged from the team’s experience working with companies on projects that required them to interact with legacy systems written in programming languages such as COBOL, Fortran and Perl—languages no longer taught in universities. The difficulties they observed led them to explore the broader issues companies face when maintaining these systems.

“Both of us have been a part of the App Dev club,” Vinchhi said. “We’ve been running it for a year and have done contracting work for all these big companies. When we brought in new students, they were confused by the legacy technologies many large companies use. This experience showed us firsthand how legacy systems can pose a significant challenge for businesses onboarding new talent.”

Their exposure to the problem prompted the team to attend conferences in Northern Virginia, where they engaged with industry professionals and gained further insight into the challenges surrounding legacy code. They soon realized that while companies recognize the need to modernize these systems, the process is often prohibitively expensive and time-consuming.

“There’s a real business opportunity here,” Vinchhi said. “We’ve been building tools and participating in hackathons for a long time, and we saw a chance to create something that addresses a major pain point in the industry.”

Mission and Approach  

Axal’s mission is to streamline the modernization of legacy code through its Software-as-a-Service (SaaS) tool. The tool automates version updates, code migration and ETL development using custom agents. The platform helps companies seamlessly transition outdated systems to a modernized architecture.

“Axal is an AI-based platform,” Patel said. “Right now, we are using the tool for manual migrations because AI is not 100% accurate, and there needs to be a human overseeing the process. The goal is to enable autonomous migration to any platform without needing review.”

Legacy code presents multiple challenges, especially when companies struggle to hire new developers for these outdated languages. Such systems often require senior developers already familiar with the technology, creating a costly dependency on a shrinking talent pool. 

“When you have to add new features or expand a platform, it’s not just about costs; it’s about your future competitive advantage against your competitors,” Patel added. “Modern systems allow for quicker onboarding and more seamless feature additions.”

Utilizing AI 

Axal’s technology leverages AI to break down codebases into modules and translate them into newer programming languages or architectures, reducing the time and cost required to modernize outdated systems. For example, a COBOL codebase on a mainframe could be converted into a Java microservices architecture using Axal’s tools.

The platform’s static analysis tools deconstruct a codebase into smaller components before passing them into AI models for translation. By analyzing and reconstructing code intelligently, Axal ensures the final product maintains functionality and performance, minimizing the risk of errors during migration.

“Our approach is unique because we use AI to translate parts of the code and then put it back together,” Vinchhi said. “We also generate tests automatically using our large language models to ensure the new system works as expected.”

Differentiation

According to the team, Axal is not the only company tackling legacy code modernization, but it differentiates itself from its competitors. 

“There are two types of companies in this space,” Vinchhi shared. “One group is trying to build general-purpose software agents, but we believe AI isn’t advanced enough yet for enterprises to trust and pay for these agents. The other type includes companies that do basic code translation using static analysis. Neither approach is ideal.”

Axal’s solution lies in the middle ground, using static analysis to understand code and AI to translate small chunks reliably. This combination creates a robust and dependable platform for code modernization.

“We’re the only company, as far as we know, that’s taking this specific approach,” Vinchhi noted.

Challenges and Hurdles

While Axal’s vision is clear, the road to achieving it is filled with challenges. One of the most significant technical hurdles has been developing the static analysis tools required to parse complex legacy codebases.

“Passing things into AI models and prompting isn’t too difficult,” Vinchhi explained. “The hard part is getting our indexers and parsing tools to understand these super legacy codebases. We’ve had to build a lot of custom tooling and systems to make this possible.”

Finding the right product-market fit has been another challenge on the business side. 

“This isn’t our first idea in the developer tool space,” Vinchhi said. “We’ve built a lot of things that didn’t work out. Moving slowly towards product-market fit has been tough, but we believe we’re getting closer.”

Hatchery’s Impact

Being selected for the Mokhtarzada Hatchery Program has been a significant step for Axal. The program provides mentorship and support to student entrepreneurs, helping them refine their business models and accelerate growth.

“It’s honestly an honor to be part of the Hatchery,” Patel expressed. “The direct mentorship from the Mokhtarzada brothers has been invaluable. We’ve already had sessions where they’ve given us great advice on moving forward. The accountability and support from other teams in the program have also been beneficial.”

The Hatchery program offers a collaborative environment where ideas are freely exchanged and connections are made. 

“It’s a great space for refining our product and business strategies,” Patel added.

Plans for the Future  

Axal’s immediate goal is to continue developing autonomous agents that can handle migrations and refactors autonomously.

“We’re currently focusing on converting legacy code like COBOL and Fortran into newer languages like Golang, Java and Python,” Patel said. “But in the next five years, we see a future where coding might not be the primary focus.”

The team believes systems will shift towards natural language interfaces and no-code solutions as AI technology evolves.

“It’s not just about getting companies out of the mainframe era,” Patel noted. “We want to be prepared for the next wave of modernization, where systems are accessible through natural language and no-code platforms.”

As Axal continues its journey within the Mokhtarzada Hatchery Program, it remains focused on its mission to make legacy code modernization more accessible and efficient. With AI-driven tools and a dedicated team, the startup aims to redefine how businesses approach legacy systems and prepare for a future of innovation and growth.

—Story by Samuel Malede Zewdu, CS Communications 

The Department welcomes comments, suggestions and corrections.  Send email to editor@cs.umd.edu.

Read the article here.