.png){kind=logo}
Quantum error correction is usually described as finding a broken qubit and fixing it. Vedika Saravanan's work highlights the real constraint: a stopwatch. In some quantum computing architectures, the processor keeps computing while the classical machine works out what went wrong.
The standard account of quantum error correction makes it sound like proofreading. The machine notices a qubit has gone wrong and corrects it. Problem identified, problem solved.
That description leaves out the part that makes it hard. The errors happen while the computation is still running. Information about them has to be pulled out of the machine without collapsing the fragile logical quantum state the computation depends on. That information then goes to a classical computer, which has to work out what most likely went wrong and decide what to do about it. And throughout all of this, the quantum processor does not necessarily wait.
"The difficult part is that the quantum processor does not stop while the classical system thinks," says Vedika Saravanan, whose doctoral research focused on quantum error mitigation and reliability, and whose current work extends into quantum error correction and classical decoding. She calls it "the race between error generation and error interpretation", and she considers it one of the most important engineering problems standing between today's machines and fault-tolerant ones.
Saravanan holds a PhD in electrical engineering, with doctoral research centred on scalable quantum compilation, error mitigation, noise-aware circuit execution and reliability modelling for noisy quantum systems. She has more than 11 peer-reviewed publications across quantum computing, hardware reliability and security, in venues including IEEE Transactions on Computer-Aided Design, IEEE Transactions on Quantum Engineering, the IEEE European Test Symposium and the International Symposium on Quality Electronic Design, and has contributed to quantum physics work published in Physical Review Letters.
The reason the stopwatch matters is that quantum errors do not sit still. "A small error introduced early in a circuit can propagate through subsequent gates and affect multiple qubits," she says. It is one of the challenges that separates quantum error management from conventional fault detection. In quantum systems, errors can propagate through subsequent operations, making their effects increasingly complex to interpret.
So a decoder that is accurate but slow may still create a serious bottleneck. Although some corrections can be tracked and applied later, the classical system must keep pace with the incoming error information and provide results in time for operations that require feedback.
Her research has examined quantum reliability across several interconnected layers. In compilation, she worked on mapping circuits onto the more reliable regions of a processor, treating device-level noise as an input to the decision rather than assuming every qubit and coupling behaves identically. That work used reinforcement learning and graph neural networks to find placements intended to improve execution reliability.
In reliability modelling, she developed data-driven methods to predict which circuits and which mappings were most vulnerable to hardware noise, using machine learning and graph representations of the circuits themselves. The point was not prediction for its own sake but to give compilers and runtime systems something concrete to act on.
In error mitigation, she analyzed how Pauli errors propagate through a circuit and used that information to reorganize the order in which gates execute, with the aim of reducing the effects of accumulated errors without requiring full quantum error correction. She has also worked on the security side of the same hardware behavior, including decomposition-based watermarking of circuits and attacks that exploit the error characteristics of noisy devices, on the grounds that imperfections affect not only whether a computation is right but whether it can be trusted.
Her professional work has since expanded into quantum error correction and classical decoding, where the challenge extends beyond mitigating noise to interpreting error information efficiently as quantum computations proceed.
A central observation from her research is that theoretically equivalent quantum circuits can behave differently when executed on physical hardware. "Two theoretically equivalent circuit implementations can behave very differently on a real noisy device," she says, "because qubits, couplers, and operations do not have identical error characteristics.” This highlights why hardware aware compilation and reliability analysis are essential to practical quantum computing.
Which brings her to the conclusion she states most firmly, and the one that cuts against how progress in the field is usually reported. "Quantum reliability cannot be solved at only one layer of the computing stack," she says. "Better qubits alone are not enough."
It is a pointed thing to say in an industry that announces qubit counts. Her position is that reliable quantum computation depends simultaneously on compilation, control, error detection, decoding, scheduling and the interaction between the classical and quantum halves of the machine, and that optimizing any one of them in isolation runs into the others. Devices, error-correcting codes, decoders, compilers, control electronics and software, in her account, cannot be designed independently.
It also reframes what scaling means. The issue is not only how many physical qubits can be manufactured, but whether the classical infrastructure around them can process error information at the rate the quantum hardware generates it. A larger processor can produce more syndrome data with each measurement cycle. If decoding cannot keep up, more qubits make the race harder to win rather than easier.
The shift she is really arguing for is conceptual. A quantum computer, in her description, should not be pictured as a quantum processor with a classical computer attached to it. It is more accurately a tightly coupled system in which quantum computation and classical error-processing operate together, with overall performance shaped by their timing, communication and reliability requirements.
The race is not simply about building quantum processors with more qubits. It is about ensuring that the systems responsible for detecting, interpreting and managing errors can keep pace with the computation itself. For Saravanan, the path to reliable quantum computing lies in bringing hardware, software and classical error-processing together, so that progress in quantum computing is measured not only by how much a machine can compute, but by how reliably it can do so.
