Quantum risk models
Qiskit circuits for cybersecurity risk, including ancilla circuits that simulate non-unitary operations.
- Role
- Quantum Software Engineering Intern
- When
- Oct 2024 – Apr 2025
- Where
- Buenos Aires
- Stack
90%+
accuracy against classical methods
up to 30%
improvement in algorithm efficiency
Context
A four-student research team at UAI in Buenos Aires, supervised by faculty there, working on quantum software engineering for risk: how failures cascade through a network when each link fails with some probability.
What I built
- Risk models in Qiskit, run on simulated IBM Quantum backends, with circuits tuned for the noise limits of current hardware. Algorithm efficiency improved by up to 30%.
- Ancilla circuits for non-unitary operations. Quantum gates are reversible, but risk models need operations that are not. Adding ancilla qubits lets a circuit simulate them; ours matched classical methods at 90%+ accuracy.
- A simulation framework for cascading failures in probabilistic networks, used to compare the quantum results with classical Monte Carlo runs.