The Midnight Phone Call
There is a profound difference between the threats listed neatly in a corporate risk register and the actual problems that wake an operations director at three in the morning. Most massive projects fail because leaders accept a baseline they should challenge. Muriel Demarcus knows this reality intimately. She spent three decades delivering billion-dollar infrastructure before realizing the real threat was not the engineering. The threat was the silence before the failure.
The Engineer Who Found the Blind Spot
Muriel Demarcus is the CEO of Marsham Edge. She spent thirty years directing commercial and major projects for organizations like Sydney Metro and London Underground. Today, she builds artificial intelligence systems that detect critical infrastructure failures before the control room even knows a crisis is brewing.
Decades Spent Building Systems That Cannot Fail
Her understanding of risk did not come from a textbook. It came from the trenches of high-stakes public transport and energy delivery. She learned early on that large, consequential projects demand absolute precision. A system that mostly works is actually a system waiting to break.
Her technical foundation started with Fortran punch cards and an applied mathematics degree from Ecole Centrale de Paris. But her real education happened on the ground. During her tenure as Principal at London Underground, she faced a massive operational challenge. She brought maintenance and upgrades back in-house through intense contract renegotiations over a seven-year cycle.
This move dramatically improved service quality and helped secure a massive £323 million adjudication win. The experience sharpened her ability to spot structural flaws in massive contracts. Later, she took on the role of Commercial Director for Sydney Metro. She advanced the delivery of driverless train technology for Australia’s largest public transport project.
Managing a $1.6 billion package required her to look past surface-level reporting. She had to find the hidden integration risks that could derail the entire timeline. This relentless questioning of the baseline reached its peak at Transgrid. As Major Projects Director, she led a massive renewable energy zone program.
The original scope assumed expensive greenfield easements. She challenged that fundamental assumption. By redesigning the delivery around the existing network, she generated approximately $1.5 billion in savings across contract values. This bold move also accelerated the schedule by two full years.
“Most programs solve the wrong problem because the original brief remains an unexamined black box,” she says. Her ability to open that box defined her corporate career. Now, she applies that exact scrutiny to machine learning.
Training Algorithms on the Physics of Failure
Today, Muriel Demarcus leads Marsham Edge. Her company builds anomaly detection systems for critical infrastructure and energy markets. They do not build theoretical models. They build systems that operate in production to stop disasters.
Her team recently tested their technology on a national grid blackout scenario. The model flagged the cascade precisely 75 seconds before over a million homes lost power. That brief window is the difference between a managed shutdown and a catastrophic failure.
But she knows that a technical warning means nothing without the right operational structure. The best algorithm in the world cannot save a grid if the human operator does not trust the alert.
“A 75-second window is only valuable if it reaches someone with the authority and protocol to act,” she explains. A raw probability score on a dashboard is useless in a crisis. Her systems deliver plain-language briefings that explain the physics of the impending event.
Beyond grid blackouts, her team tackles battery safety and predictive maintenance. They audit battery management system data using machine learning classification. This process flags thermal runaway risks in lithium-ion systems long before a fire starts. They also forecast energy demand and predict prices for massive consumers. Every product they build is rooted in the physical realities of heavy industry.
She built this pragmatic approach after watching countless technology deployments fail. She noticed a pattern in how executives reviewed progress. She refused to accept status reports that only showed the good news.
“Percentage-complete is a vanity metric if the remaining 20 percent contains the entirety of the integration risk,” she notes. She pushed accountability down to the data originators. She made it structurally safe for her teams to surface bad news early.
At Marsham Edge, she applies this same operational rigor to artificial intelligence. She sees artificial intelligence programs die quietly because organizations lack governance. They have the data but no clarity on who owns the output.
“The 3 a.m. crises are almost always issues that were visible at 3 p.m. six months prior,” she says. Her current work ensures those early signals are never ignored again.
The Signal in the Noise
The infrastructure we rely on every day is growing more complex and more fragile. The old methods of risk management are no longer sufficient to prevent catastrophic failures. Muriel Demarcus spent her career learning exactly how these massive systems break under pressure.
Now, she gives operators the tools to see the breaking point before it arrives. She bridges the massive gap between civil engineering and artificial intelligence deployment. The midnight phone call does not have to ring.
Muriel Demarcus is the CEO of Marsham Edge based in Singapore. She builds artificial intelligence anomaly detection systems for critical infrastructure and energy operators. To connect with Muriel or learn more, visit her LinkedIn profile.
Muriel re-write:
The Brief No One Questions, the AI No One Uses: Muriel Demarcus on Fixing How Big Infrastructure Gets Risk Wrong
The Midnight Phone Call
There is a profound difference between the threats listed neatly in a corporate risk register and the ones that wake an operations director at three in the morning. Most large projects fail because leaders accept a baseline they should have challenged. Muriel Demarcus knows this reality intimately. She spent three decades delivering billion-dollar infrastructure before realising the real threat was not the engineering. The threat was the silence before the failure.
The Engineer Who Found the Blind Spot
Muriel Demarcus is the CEO of Marsham Edge. She spent thirty years directing commercial and major projects for organisations like Sydney Metro and London Underground. Today, she builds artificial intelligence systems that detect critical infrastructure failures before the control room knows a crisis is coming.
Her understanding of risk did not come from a textbook. It came from the trenches of high-stakes public transport and energy delivery. She learned early that large, consequential projects demand absolute precision. A system that mostly works is a system waiting to break.
Her technical foundation began with Fortran punch cards and an applied mathematics degree from École Centrale de Paris. Her real education happened on the ground. As Principal at London Underground, she faced a significant operational challenge: maintenance and upgrades had been fragmented across external contracts. She renegotiated and the contract was eventually brought back in-house, a move that improved service quality. Along the way her team secured a £323 million adjudication win. The experience sharpened her ability to spot structural flaws buried deep in complex contracts.
Later, as Commercial Director for Sydney Metro, she advanced the delivery of driverless train technology for Australia’s largest public transport project. Managing a $1.6 billion package required her to look past surface-level reporting and find the hidden integration risks that could derail the entire programme.
This relentless questioning of the baseline reached its peak at Transgrid. As Major Projects Director, she led a renewable energy zone programme whose original scope assumed expensive greenfield easements. She challenged that assumption. By redesigning delivery around the existing network, she generated massive savings and accelerated the delivery schedule..
“Most programmes solve the wrong problem because the original brief remains an unexamined black box,” she says. Her ability to open that box defined her corporate career. Now she applies that same scrutiny to machine learning.
Training Algorithms on the Physics of Failure
Marsham Edge builds anomaly detection systems for critical infrastructure and energy markets. Not theoretical models, systems operating in production to stop disasters.
The team tested their technology against a national grid blackout scenario. Their model flagged the cascade precisely 75 seconds before over a million homes lost power. That window is the difference between a managed shutdown and a catastrophic failure.
But Muriel knows a technical warning means nothing without the right operational structure. “A 75-second window is only valuable if it reaches someone with the authority and protocol to act,” she says. Raw probability scores on a dashboard are useless in a crisis. Her systems deliver plain-language briefings that explain the physics of the impending event.
Beyond grid blackouts, the platform addresses battery safety and predictive maintenance — auditing battery management system data to flag thermal runaway risks in lithium-ion cells long before a fire starts, and forecasting energy demand and prices for large-scale consumers. Every product is rooted in the physical realities of heavy industry.
This approach earned international recognition in June 2026, when Marsham Edge won first place at the Epic Hackathon in Singapore, one of the region’s most competitive applied AI challenges, with judges drawn from defence, energy and technology. The win validated what Muriel had always argued: that an AI platform built on operational rigour, not theoretical elegance, is what the real world needs.
The Signal in the Noise
She built this pragmatic approach after watching technology deployments fail in a recurring pattern. Executives reviewed progress through status reports that only surfaced good news.
“Percentage-complete is a vanity metric if the remaining 20 percent contains the entirety of the integration risk,” she notes. She pushed accountability down to the data originators and made it structurally safe for teams to surface bad news early. At Marsham Edge, she applies that same governance discipline to artificial intelligence — because AI programmes fail quietly when organisations have the data but no clarity on who owns the output.
“The 3 a.m. crises are almost always issues that were visible at 3 p.m. six months prior,” she says. Her work ensures those early signals are never ignored again.
The infrastructure we depend on is growing more complex and more fragile. The old methods of risk management can no longer prevent catastrophic failures. Muriel Demarcus spent her career learning exactly how these systems break under pressure. Now she gives operators the tools to see the breaking point before it arrives — and the midnight phone call doesn’t have to ring.
Muriel Demarcus is CEO of Marsham Edge, Singapore. She builds AI anomaly detection systems for critical infrastructure and energy operators. Connect with her on LinkedIn at https://www.linkedin.com/in/muriel-demarcus-a8165b2/ or visit https://marshamedge.com./
