After 25 years moving technology from test benches into mass production, the CEO of QubitNexus.AI is building software that makes quantum computers of every kind more reliable, comparable and usable by people who are not physicists.
Prof. Dr. Kathrin Kind’s career began in 1999 at Siemens, in quality engineering for international mobile networks. The work came down to one question, and she has been answering versions of it ever since: will a technology that behaves in the lab keep behaving once millions of people depend on it?
In the automotive industry, the same question carried higher stakes. She went on to develop technology for Bosch, ZF, Magneti Marelli, BMW, Mercedes-Benz, Audi and the Volkswagen Group, working on engine management, braking, steering, cockpit infotainment, driver assistance and autonomous driving platforms. A braking system that passes on the test rig still has to work in every car that leaves the factory. Later she became Chief Data Scientist and AI Director at Cognizant, where she helped large organizations put AI into production. By her count, she has industrialized more than 200 software platforms across 20 industries for large enterprise clients.
Quantum computing is her latest test case. The field has produced plenty of laboratory records and very few systems an enterprise can run with confidence. Her company, QubitNexus.AI, writes software that works across quantum machines of every type, and its case to industry rests on reliability.
Errors Decide Who Reaches Industry
Quantum computers store information in qubits, which can hold a blend of 0 and 1 and can be linked to one another. In principle, that lets them solve certain problems in chemistry, optimization and AI that overwhelm ordinary computers. Qubits are also fragile. Heat, vibration and stray signals knock them off course, and in 2018 the physicist John Preskill warned that noise in quantum gates would limit how large a circuit could run reliably. QubitNexus.AI sees errors as the main barrier between quantum computing and industrial use.
The hardware market adds a second problem. An enterprise can rent time on superconducting machines chilled to a fraction of a degree above absolute zero, on neutral-atom systems that hold individual atoms in laser beams, on photonic processors that compute with particles of light, or on classical simulators. Each type has its own error patterns and output formats, so a result from one machine is hard to compare with a result from another. QubitNexus.AI was designed to address both problems without tying itself to any single kind of hardware.
A Language That Rejects Risky Code
The company’s approach starts with a programming language. Q11L™, which Dr. Kind invented, is proprietary and comes with its own compiler, the program that turns code into instructions a machine can execute.
Its central idea rests on a basic rule of quantum physics. A qubit cannot be copied, and once it is measured, the information it held is gone. Code that treats a qubit as reusable will fail, often after it has already consumed expensive machine time. Q11L uses what computer scientists call affine types, a rule that each value may be used at most once, so the compiler rejects such programs before a job reaches a quantum computer. The error gets caught at compile time, on the developer’s own machine.
In simulations on NVIDIA’s platform, Q11L showed higher reliability than conventional layered software frameworks and improved error correction. The results come from simulation and have yet to be confirmed on physical quantum hardware. Working in Q11L, the team has also developed proprietary algorithms that are now built into the company’s software stack.
One Yardstick for Many Machines
QubitNexus.AI’s main differentiator sits on top of that reliability work. Its unified platform sends the same job to several different quantum computers at once, collects the outputs, normalizes them and reports two things for each machine: the outcome and a reliability score on a common scale.
For an enterprise, that answers two questions it otherwise has to guess at. Which machine produced the better answer, and how far should the result be trusted? Superconducting, photonic and neutral-atom hardware can be judged side by side, along with classical simulators. A buyer gets a measured basis for deciding where to run a workload.
Quantum for People Who Are Not Physicists
Quantum hardware also needs constant tuning, known as calibration, and that job has traditionally fallen to specialists. TongueChow, the company’s patent-pending calibration and AI engine, is built to take that burden off users who are not quantum experts. QubitNexus.AI describes it as a practical interface to quantum computing designed for laypeople, the kind of tool an analyst or engineering manager could use without a physics degree.
Q11L and TongueChow sit within a full software stack. Beneath the language, the QNAI operating system and runtime translate Q11L into physical operations on whichever machine runs the job. A data management framework records where quantum-AI data came from and checks its quality. QNAI Synergy carries the proprietary algorithms. It is a hypervisor, which in ordinary computing is the layer that lets software run independently of the physical machine underneath. Synergy-QNAI™ runs today on NVIDIA GPUs and is designed to target both cryogenically cooled quantum computers and room-temperature hardware. The team builds and simulates with NVIDIA’s CUDA-Q and plans to scale simulation and hybrid quantum-classical training with cuQuantum.
The breadth serves the company’s mission of carbon-neutral intelligence, quantum AI that is “faster, leaner, and greener.” Catching errors before a job runs is designed to save machine time and the energy behind it, and support for room-temperature machines gives customers a route that avoids the power cost of deep cooling. The core products, including Q11L™, TongueChow and QNAI Synergy, are protected by trademarks and pending patents.
The People Behind It
Dr. Kind teaches as well as builds. She is Professor of Artificial Intelligence and Quantum Computing at the Paris School of Management and at SSBM Geneva, and a technical editor for Springer Nature and CRC Press. She holds a doctorate in AI applied to business, an MBA in artificial intelligence and a master’s degree in software engineering.
Her governance work runs in parallel. She is a member of the World Economic Forum’s Global Future Council on Data Frontiers, the Responsible AI Governor for Switzerland at the Global Council for Responsible AI, and an Honorary Ambassador for the UN Support Action Centre (JDI). QubitNexus.AI takes part in UN digital cooperation work and related responsible-AI and digital-governance initiatives. At the company, responsible AI includes counting the carbon.
Around her is a team with more than 400 years of combined experience, including physicists, engineers and quantum architects from MIT, NUS, TU Wien, Cambridge, Columbia and ETH Zurich, alongside operators from large corporations. Its board also includes an AI, which holds the title of Chief Marketing Officer.
QubitNexus.AI has funded itself from the start. It spent its first year on pure research, and in its second it signed eleven paying enterprise and institutional customers on advisory and training contracts, with more enterprise deals in its pipeline. Its services also include consulting on the move to post-quantum cryptography and quantum training for executives and engineers.
Making Quantum Measurable
Quantum computing has spent years setting records under laboratory conditions. The harder test is the one Dr. Kind has faced in networks, cars and enterprise AI: whether the people who depend on a technology can measure it and trust it, at an energy cost a responsible business can defend. QubitNexus.AI is betting that software built for every kind of quantum machine can pass that test at enterprise scale.
Prof. Dr. Kathrin G. Kind is Global CEO of QubitNexus.AI and Professor of Artificial Intelligence and Quantum Computing at the Paris School of Management and SSBM Geneva. She is a member of the World Economic Forum’s Global Future Council on Data Frontiers. qubitnexus.ai
