Monday, August 3, 2026

Compute & Cloud

MayimFlow uses edge AI to predict data center water leaks

MayimFlow uses IoT sensors and edge-deployed machine learning to potentially provide 24 to 48 hours of warning before damaging water leaks occur in data centers.

MayimFlow uses edge AI to predict data center water leaks

As data centers expand to support computing demands, managing physical infrastructure risks has become a critical operational challenge. MayimFlow, a startup that was a Built World stage winner at the TechCrunch Disrupt event, is focusing on preventing damaging water leaks in these facilities. Currently, many data centers rely only on reactive solutions for water leaks, according to MayimFlow founder John Khazraee. When leaks occur, they can cause significant operational downtime and financial loss for operators.

To address this, MayimFlow has developed a system that combines Internet of Things (IoT) sensors used for monitoring with edge-deployed machine learning models—which are AI models running locally on hardware rather than in the cloud—to detect signs of impending leaks. Khazraee, who spent more than 15 years building infrastructure for technology companies including IBM, Oracle, and Microsoft, designed the system after observing these vulnerabilities firsthand. According to John Khazraee, founder of MayimFlow, “I’ve noticed these issues in data centers, and the only solution they had was: ‘When the leak happens, we find out,’” He noted that reactive approaches require significant remediation spending, server shutdowns, and data disruptions, which motivated him to build a preventative solution. He believes the system can provide 24 to 48 hours of advanced warning, allowing operators to make repairs before critical systems fail.

Khazraee’s focus on resource efficiency is rooted in his childhood. Growing up in a family that was not highly well-off, his father would often tell him he was in the shower too long and ask if he was singing inside. This early emphasis on conservation shaped his approach to engineering and infrastructure. Over the last two years, Khazraee has turned down roles at multiple Big Tech companies to build MayimFlow. While the company is currently targeting data centers, Khazraee wants to bring this solution beyond data centers to commercial buildings, hospitals, manufacturing facilities, and possibly even utilities. Khazraee, who believes deeply in the company’s vision and impact, noted that water is becoming one of the major issues in our world.

Why it matters

MayimFlow is attempting to address the significant financial and operational risks posed by water leaks in data centers by providing predictive monitoring solutions. As infrastructure demands grow, shifting from reactive to predictive maintenance is crucial for preventing costly downtime.