Aleix Moreno Telesforo

Screened 6,232 records down to a 777-paper corpus, classified against a taxonomy we built rather than borrowed, and synthesised across eight problem areas.

Stress-tested the 13 implementations the literature nominates as most likely to show an advantage, using Microsoft's Azure Quantum Resource Estimator — and counted the cost most papers leave out: loading the classical data into the quantum machine.

With that cost included, none of the 13 clears the bar. The bottleneck is the classical–quantum interface, which reframes quantum finance as an integration problem rather than a shopping list of algorithms.

Method: corpus and codebook frozen before analysis, success criteria pre-registered, and every analytical theme re-checked by a second language model from a different family — which demoted 11 of 45 themes as unsupported.

Supervisor: Prof. Raghava Rao Mukkamala. Co-author: Vincent Wallerich. Microsoft supported the project with model and compute access.

I would call that method a defensible design pattern for research in fast-moving technical fields — not a validated standard, which is not a claim one thesis can support.

Audit circularization — writing to a client's counterparties to confirm balances — is slow, manual and trust-dependent. We asked whether triple-entry accounting could do it structurally instead: the blockchain as the third book, the one both parties can see.

A venture plan, not a paper. Market and competitor analysis, product and pricing, operations, hiring, legal and fiscal structure, and a financial feasibility model through pre-seed and seed.

Conclusion: real market, real margins, no dominant player yet — and cost and regulatory risk that external financing would have to carry.

Written with Gerard Merino Lago. The first time I did the thing I keep doing: take a technology people are excited about and ask what it would take to actually work in a business.

Next.js 16, React 19, TypeScript and CSS Modules. Statically prerendered, with no client-side JavaScript beyond a single disclosure primitive.

Product brief, technical design, content spec and decision log live in the repository, versioned with the code that implements them.

If I am going to claim I can build and evaluate AI systems, the site making the claim may as well be one.

What actually changes when the model runs on your desk instead of someone else's cloud.