Portfolio
Projects
Academic, hackathon, and hobby projects spanning quantum computing, quantum information science, and machine learning.
MM-HSD: Multi-Modal Hate Speech Detection in Videos
Published at ACM Multimedia 2025 (originating as an EPFL Deep Learning course project, selected from 50 groups for publication support). A multi-modal model integrating video, audio and text via Cross-Modal Attention; reaches state-of-the-art M-F1 0.874 on the HateMM dataset.
Fine-Tuned RNA Language Models for Branch Point Prediction
Research paper accepted at the ICLR 2025 MLGenX and AI4NA workshops (EPFL Machine Learning course). Fine-tuned and evaluated multiple RNA language models for intronic branch-point prediction, achieving state-of-the-art performance.
Circuit Knitting and Tensor Networks for Quantum Simulation
Semester project at the Computational Quantum Science Lab, EPFL (supervised by Prof. Giuseppe Carleo). Integrated circuit knitting and tensor-network methods to tackle the simulation of strongly correlated quantum systems on NISQ devices.
Machine Learning for Wigner Negativity Classification
Summer research at the Quantum Information and Computation Lab, EPFL (supervised by Prof. Zoë Holmes). Developed ML approaches (Neural Networks, Gaussian Processes) to classify Wigner negativity in continuous-variable quantum systems, reaching up to 95% accuracy on cat and coherent states.
Computational Analysis of Matchgate Circuits
Semester project at SMILS Lab, EPFL (supervised by Prof. Nicolas Macris). Analysed the computational complexity of classically simulating matchgate circuits and evaluated techniques to simulate fermionic quadratic Hamiltonians in polynomial time.
ETH Quantum Hackathon: Quantinuum Challenge
Achieved 2nd place implementing the Heyfron-Campbell "normal form" of arbitrary quantum circuits, a technique useful for reducing T-gate count in quantum compilation.
AlHaQ: NYU Abu Dhabi Quantum Hackathon
Built a multi-modal ML model using Quantum Convolutional Neural Networks to classify fake-news tweets, and a QUBO model to rank Twitter feeds by "trustedness". Received a travel grant from EPFL's Quantum Science and Engineering Center.
Noise Thresholds for Classical Simulation of Gaussian Boson Sampling
BSc end-of-degree project at Imperial College London (supervised by Prof. Myungshik Kim). Determined, analytically and numerically, the noise thresholds for efficient classical simulation of Gaussian Boson Sampling, a candidate route to photonic quantum advantage.