Projects.

The AI department built from scratch: an enterprise-intelligence ecosystem in the spirit of Palantir —a company’s data unified and actionable, with AI reasoning on top— deployed entirely on in-house infrastructure. The direction is clear: from Retrieval-Augmented Generation (RAG) to Ontology-Augmented Generation, from the retrieved document to knowledge structured in a company model.

Scope

• AI strategy and priority use case, defended before the leadership.
• End-to-end RAG architecture: ingestion, chunking, vector indexing and retrieval.
• In-house infrastructure: not a single datum goes to third parties.
• Roadmap toward a company ontology: from RAG to ontology-augmented generation.

Definition and rollout of the factory's digital backbone: ERP, network, access control and automation of administrative and production processes.

Scope

• ERP architecture and mapping of business processes.
• Access management, network segmentation and encryption.
• Attendance control through biometric security.
• Automation of repetitive tasks across systems.

A complete simulation chain on automotive components, from meshing through to the interpretation of results.

Scope

• ANSA — Preprocessing and mesh generation.
• Pam-Crash — Dynamic and impact analysis.
• Abaqus — Static and non-linear analysis.
• Meta — Postprocessing and reporting.

End-to-end management of industrial clients through the process of obtaining innovation incentives, with precise technical documentation as the foundation of every application.

Scope

• Qualifying which projects genuinely meet the criteria for R&D or technological innovation.
• Drafting the technical report that holds up before whoever assesses it.
• Processing tax deductions, rebates and grants, both self-assessed and certified.
• Running the liaison with the client and with the evaluating bodies.

The double master’s produced two complementary theses (UPC-ETSEIB, 2025), and this is the build: a complete intelligent energy management architecture, assembled and validated end to end on low-cost hardware. Current and voltage sensing, containerised microservices, time series, real-time dashboards, a local language model as the analyst and the official electricity market prices flowing in every hour. It is the same pattern I apply today at company scale: in-house infrastructure, real-time data and a layer of intelligence on top.

Scope

• A prototype validated end to end: Raspberry Pi with current and voltage sensing, Kubernetes (MicroK8s), InfluxDB and Grafana in real time.
• Official market prices (PVPC) ingested hourly from the Spanish grid operator’s API, to decide when to consume with the tariff in view.
• A local LLM deployed in the cluster itself as the analyst: consumption recommendations combining live readings and price.
• Design of the peer-to-peer energy market on Cardano: smart contracts in Plutus, decentralised identity (DIDs) and NFT traceability.

The author’s works at UPCommons (UPC) →

The second thesis of the double master’s (UPC-ETSEIB, 2025), and the business one: it adapts the same architecture to industrial settings and answers the question a prototype cannot answer alone — is there a market? An analysis of the European energy management sector, a Technology Readiness Level (TRL) comparison against the market leaders and a ten-year business plan with its projections and risks. The document is embargoed under a confidentiality agreement until 2030; this is the part that can be told.

Scope

• A study of the European home and industrial energy management market: trends, adoption barriers and competitive analysis.
• A systematic Technology Readiness Level (TRL) comparison against existing solutions, ahead on the critical functionalities.
• A ten-year business plan with multiple revenue streams — hardware, subscriptions and transaction fees — targeting 1-2% of Spain’s free energy market.
• Market, operational and financial risk analysis, each with its mitigation strategy.

Record at UPCommons (UPC) →

Bachelor’s thesis (UPC-ETSEIB, 2022): an engineering analysis of the tricarboxylic acid cycle — the metabolic pathway every living cell uses to produce its energy — as a clean energy source. The result is the complete design of a harvester fed by the blood’s own glucose, sized to power a pacemaker for the patient’s entire life, and for which I began the patent filing process. It was the first time I treated a living system as an engineering system; it was not the last.

Scope

• A complete-oxidation biofuel cell: the full TCA enzymatic cascade mounted on the anode. Conventional glucose cells extract 2 of the 24 available electrons — 8.3% of the energy; this design goes after all of them.
• ANSYS CFD simulation of the arterial bifurcation: blood as a non-Newtonian fluid (Carreau model) with pulsatile flow, verifying the tap does not damage the artery.
• Sized for a lifetime pacemaker with a 10 mm² anode: daily consumption amounts to 0.018% of the body’s glucose.
• Alternative selected through multi-criteria analysis (AHP), economic evaluation and patent filing initiated; document access restricted by the author’s decision.

Record at UPCommons (UPC) →