012025
Train on classical, deploy on quantum: scaling generative quantum machine learning to a thousand qubits
Training large quantum generative models on classical hardware for deployment on quantum devices.
arXiv:2503.02934 · 2025
Research scientist · Quantum & Classical Machine Learning
Thinking from first principles.
Learning from experiments.
I’m Erik, a researcher working at the intersection of machine learning and quantum computing. I study how learning models work, where their limits lie, and how quantum computers could surpass the capabilities of the AI models we currently have. My true passion is understanding from the ground up, so acting from first principles is in my DNA.
Writing my PhD thesis at ICFO · Expected completion before summer 2027
Learning theory, efficient inference,
and scalable generative models.
022025
IQPopt: Fast optimization of instantaneous quantum polynomial circuits in JAX
Software for optimizing large-scale quantum circuits with automatic differentiation in JAX.
arXiv:2501.04776 · 2025
032025
Single-shot quantum machine learning
Investigating when quantum classifiers can make reliable predictions with a single measurement.
Physical Review A 111, 042420 · 2025
042026
IQP Born Machines under Data-dependent and Agnostic Initialization Strategies
Understanding how initialization impacts the trainability of quantum generative models and the mitigation of barren plateaus.
arXiv:2603.14576 · 2026
052025
Potential and limitations of random Fourier features for dequantizing quantum machine learning
Characterizing when classical random Fourier features can efficiently reproduce quantum learning models.
Quantum 9, 1640 · 2025
062026
Learning complexity gradually in quantum machine learning models
A curriculum-inspired approach to training quantum models by prioritizing informative data.
Physical Review Research 8, 033006 · 2026
You may also find my work published under Erik Recio-Armengol.
I’m finishing my PhD in Quantum Machine Learning at ICFO. My work combines mathematical analysis and numerical experiments to understand learning in high-dimensional systems, with a focus on model design, efficient inference, and optimization.
I enjoy getting to the foundations of a problem: understanding why a method works, identifying its limits, and turning that understanding into useful software. Python, JAX, and PyTorch are central to my day-to-day work.
My experience spans academic research and industry, including research stays and internships at Freie Universität Berlin, Xanadu, Los Alamos National Laboratory, and Allos AI. Before research, I spent several years building software and data analysis tools.
I’m interested in Research Scientist roles where I can bring this mix of theory and implementation to fundamental questions in AI.
03 / Say hello
Let’s work on interesting problems.
Research collaborations, ideas, or opportunities in machine learning and quantum computing.
erik@armengol.me