Internship Highlight: Vani Nigam, SMS group

At the intersection of artificial intelligence (AI) systems and materials processing, materials science and engineering graduate student Vani Nigam turned academic theory into real-world impact in her role as a Data Science Intern with SMS Group this summer.  

Her internship focused on the development of a modular multi-agent system designed to streamline root cause analysis and delay prediction within steel production environments. By integrating AI agents built with AgentOS with Model Context Protocol (MCP) data sources, Vani helped her colleagues analyze complex operational bottlenecks across critical infrastructure, including hot mills, casters, and auxiliary systems.

“AI systems and materials processing are central to my academic focus,” she noted. “This experience offered the chance to further understand agent-based architectures, operating with real production data, rather than simulated data sets.”

After her research experience at Carnegie Mellon, Vani was eager to explore machine learning-based systems that were being trained from actual data recorded from the steel mills and casters. The most rewarding aspect of the experience for her was debugging the agentic framework to solve the question of how to build reliable, interpretable multi-agent pipelines in production.

 

I've gained critical insight into how computational tools are actually deployed in materials processing industries

Vani Nigam, MS MSE AI Master's Degree Student, Carnegie Mellon University

“Implementing human-in-the-loop confirmation steps and designing evaluation frameworks that surface why an agent made a particular decision has been intellectually satisfying,” she said “It's prompted me to think deeply about system reliability, and not just model accuracy.”

As Vani returns to campus to finish her MS in AI Engineering in Materials Science and Engineering, she is eager to apply her expanded knowledge, particularly in simulations of materials processing coursework, where she hopes to learn more about computational software used for materials synthesis in an industrial setting.

“I've gained critical insight into how computational tools are actually deployed in materials processing industries,” she said. “Bridging materials domain knowledge with AI integration is vital for today’s industries.”