Yale, 2011–2018 · Nature, 2019Top discovery of the year: catching and reversing a quantum jump mid-flight
For my Ph.D. with Michel Devoret at Yale, we showed that a quantum jump, random in when it starts, follows a continuous and predictable path once under way, so it can be caught and reversed mid-flight. The experiment was once considered out of reach of experimental atomic physics.
- Discover selected it as the top Math/Physical Sciences discovery in its special issue Top 50 Scientific Discoveries of the Year.
- Published in Nature as To catch and reverse a quantum jump mid-flight (2019), based on my dissertation. It is one of two of my papers in Nature’s 2025 Nobel Prize in Physics collection.
- Featured across the media worldwide, including Quanta, The Guardian, Le Monde and Wired, with live interviews on NPR.
- Yale News story of the year.
In good company: Discover's cover that year promised DNA tests, CBD and Mars.

Learning a quantum processor's noise, and canceling it
How much trustworthy science can a noisy quantum computer deliver before error correction? I started on that question when I founded the Qiskit Leap research team at IBM Quantum, and I'm still at it at Google Quantum AI. With Ewout van den Berg, Kristan Temme and colleagues, we built error-mitigation methods whose noise learning scales to large processors.
- Sparse Pauli–Lindblad noise models (Nature Physics, 2023): learn the correlated noise of a whole processor efficiently, then cancel it with probabilistic error cancellation. Used to amplify the noise instead, the same learned model became the basis of the error mitigation in IBM’s 127-qubit “utility” experiment (Nature, 2023).
- Model-free readout-error mitigation (TREX, Phys. Rev. A, 2022). Together with probabilistic error cancellation, it became a built-in option in Qiskit Runtime.
- Machine learning for error mitigation (Nature Machine Intelligence, 2024): learned models matched zero-noise extrapolation in accuracy at a fraction of the runtime cost, in experiments on up to 100 qubits.
- Now, at Google Quantum AI: error mitigation for error-corrected circuits (virtual error cancellation), from a circuit's information dynamics, and in the NMR OTOC experiment.

2013 onwardDesigning quantum chips: EPR, LOM and Qiskit Metal
Designing a superconducting quantum chip means turning a drawing into a Hamiltonian. I set out to make that path open, fast and reproducible, and the tools are below.
- The first multilayer (2.5D) architecture for superconducting circuits (Phys. Rev. Applied, 2016): the integration of a planar layout with the coherence of three-dimensional cavities.
- Energy-participation quantization (npj Quantum Information, 2021): one electromagnetic simulation gives the quantum Hamiltonian of the whole chip. I shared it as the open-source package pyEPR.
- Lumped-oscillator models (cQED with modular quasi-lumped models): a chip split into small, simulated pieces, then reassembled into one quantum circuit.
- Qiskit Metal (now Quantum Metal): I founded and led the IBM team that built the world’s first full-stack, open-source EDA software for quantum hardware (Forbes coverage). Groups around the world have used it to design and fabricate their own devices (a few are shown below).
- In 2020, MIT Technology Review named me one of its global Innovators Under 35, and I received the John Atanasoff Award from the President of Bulgaria.
2012 onwardTeaching, outreach, and a medal
I lecture at summer schools on qubits, noise and error mitigation, and I founded and hosted the IBM Qiskit Quantum Seminar (170+ episodes and 2M+ views). Back in 2012 I started Open Labs, where graduate students open pathways into science for underrepresented young scholars; it grew from Yale to Penn, Princeton, Columbia and Harvard and has reached more than 20,000 students and parents. Open Labs still runs at Yale, almost a decade and a half later. For that work I received the Yale-Jefferson Public Service Award in 2016.