About Vikranth Reddimasu

Vikranth Reddimasu is an AI engineer and M.S. Data Science graduate from the University of Maryland, College Park. He builds agentic AI systems end-to-end: the evaluation strategy, prompts and routing, model integrations, backend services, user interface, tests, and deployment. His work is grounded in a practical question: can a person use this reliably in production, not merely watch it succeed in a demo?

His projects span benchmark agents, retrieval systems, speech assessment, reinforcement learning, and distributed machine learning. Arya set a 72.0% OfficeQA accuracy record with a deliberately constrained prompt-and-tools workflow. MacFleet explores how Apple Silicon machines can cooperate on training workloads. PronounceAI combines phoneme, prosody, phrase, and learned assessment signals to make pronunciation feedback more explainable. Each project page links to inspectable source, methodology, or results where available.

Before graduate school, Vikranth founded Zero Sols while in college, grew the studio to six people, and shipped more than twenty production applications for paying clients. He also served as a Stanford University Innovation Fellow, helped run design-thinking workshops that reached more than 5,000 students, and joined the editorial board of the Change Forward Journal. Those experiences shaped how he works now: understand the user, reduce the problem to its real constraint, and ship the smallest system that can be measured honestly.