Research — In formulation
AI + quantum computing, applied to drug discovery.
Ambition is backed by a small, finished experiment, not by the vision. That is why this page is honest about its status: literature review, not executed simulations.
The question I pursue.
How do machine learning and quantum computing combine into a system capable of proposing and evaluating chemical compounds with medical utility?
It is too big for a thesis —as usual when starting out—, so the defensible version narrows it down: in which concrete drug-discovery tasks do hybrid quantum-classical methods offer a measurable advantage over classical methods, under the constraints of today's NISQ hardware? Measurable, bounded, and it admits a negative result.
- Status — nowLiterature reviewreading the anchor paper + scoping
- NextToy VQEH₂ + a classical baseline on the same dataset
- ProjectionThesis → PhDwith a public logbook in the Notebook
Reference readings
- i.Quantum Machine Learning in Drug DiscoveryChemical Reviews · 2025Anchor paper: quantum neural networks, data encoding, variational circuits, molecular property prediction and molecule generation.
- ii.Quantum-machine-assisted drug discoverynpj Drug Discovery (Nature) · 2025Where the field actually stands and what is applicable today.
- iii.Drug design on quantum computersarXiv:2301.04114Which drug-design tasks are real candidates for quantum computing and which are not.
- iv.SHARC-VQE: Simplified Hamiltonian Approach for Molecular SimulationarXiv:2407.12305An applied entry point to VQE, the hybrid quantum-classical algorithm at the core of the question.
- v.Quantum mechanics in drug design: progress, challenges, future frontiersPMC · 2020–2025 reviewState of the art drawn from 156 articles and 42 reviews.
- vi.Assessing the Benefits and Risks of Quantum ComputersarXiv:2401.16317To stay skeptical —a personal trademark— and not fall for the hype.
These are readings that guide the research, not publications of my own. When the first experiment runs, it will appear in the Notebook.
Applied studies
- 01Aeronautical-medical RAG engineapplied researchRetrieval with mandatory citation over DGAC/OACI regulations, the open engine behind the pilot fitness assessment (AeroFit).
- 02BMTSP optimizationexperimental studyComparison of four approaches (MIP in AMPL/CPLEX, LKH-3, a custom Artificial Immune System and NN+2-Opt) on TSPLIB instances.
- 03FoodNet — custom CNN vs. transfer learningtechnical reportBinary classification on Food-101 with a CNN built from scratch, a justified comparison against MobileNetV2 and a reflection on dataset biases.
Interested in collaborating on research or a graduate program? Get in touch.