About
What started back in high school as a simple experiment has grown into my true vocation. Today, I focus on designing and building scalable frontend systems at the intersection of software architecture and artificial intelligence engineering, combining thoughtful design with sound engineering practices.
I am currently working as a Frontend Engineer at Infoport Valencia, where I contribute to the development of digital solutions that optimize port logistics operations. My role involves designing accessible, scalable, and user-centered interfaces with Angular and TypeScript, supported by technologies such as SCSS, RxJS, and custom component libraries. I also integrate AI-assisted development tools into the workflow to accelerate development, refactoring, and code review tasks while maintaining high quality standards.
Between 2023 and 2025, prior to joining Infoport, I dedicated a period to applied research and independent development focused on the convergence between frontend architecture and AI engineering. During this phase I designed and prototyped four open-source solutions: Attention Matrix Explorer — interactive visualization of transformer attention matrices with D3.js and Web Workers; Semantic Search Engine — corporate semantic search engine with PostgreSQL + pgvector and HNSW indexing; Cognitive Agent Orchestrator — cognitive agent orchestrator with ReAct pattern and real-time SSE streaming; and Token Cost Analyzer — 100% client-side LLM prompt cost and tokenization analyzer with WebAssembly.
In 2026, I extended this convergence further with two new open-source projects: prompt-forge — an agentic infrastructure toolkit that gives AI coding agents persistent memory across sessions, designed to solve the statelessness problem that every AI-assisted developer faces; and Candle AI — a professional-grade technical analysis dashboard with a deterministic Bayesian confluence engine and an optional LLM narrative layer, built on top of prompt-forge as the capstone project validating the infrastructure. Together, they represent the full spectrum of my work: from developer tooling infrastructure to end-user applications, all at the intersection of frontend architecture and AI engineering.
Previously, I was part of NTTData as a Full Stack Developer intern, where I worked on the design and implementation of an employee management platform with Angular and Java. I also interned as a data analyst at Sothis, where I contributed to database analysis and design for compensation plans.
Outside of coding, you will often find me working out at the gym, hiking in the mountains, reading novels, or enjoying good company at a bar with friends.
Professional Experience
Sept 2025 — Present
Indefinite contract — Full time
Frontend Engineer · Infoport
I contribute to the development of digital solutions that optimize port logistics operations, building accessible, scalable, and user-centered interfaces with Angular and TypeScript, supported by technologies such as SCSS, RxJS, and custom component libraries. I integrate AI-assisted development tools into the workflow to accelerate development, refactoring, and code review tasks, maintaining high quality standards. I work in an agile environment, collaborating closely with analysts and backend developers to deliver robust, high-value solutions for the sector.
- Angular
- TypeScript
- SCSS
- RxJS
- Signals
- Kendo UI
- Web Components
- AI-Assisted Development
- Scrum
- Jira
- Web accessibility
Jul 2023 — Sept 2025
Applied research and independent development
Independent Software Engineer — AI & Frontend R&D · Open Source Projects
Applied research and independent development period focused on the convergence between frontend systems architecture and artificial intelligence engineering. I designed and prototyped four open-source solutions: Attention Matrix Explorer (2023) — interactive visualization of transformer attention matrices with D3.js and Web Workers; Semantic Search Engine (2024) — corporate semantic search engine with PostgreSQL + pgvector and HNSW indexing; Cognitive Agent Orchestrator (2024-2025) — cognitive agent orchestrator with ReAct pattern and real-time streaming via SSE; Token Cost Analyzer (2025) — LLM prompt cost and tokenization analyzer with tiktoken WebAssembly. Stack: Angular (16 → 19), TypeScript, RxJS, Signals, D3.js, Spring Boot, PostgreSQL, pgvector, Node.js, Express, Web Workers, WebAssembly, SSE, Docker, Ollama.
- Angular (16 → 19)
- TypeScript
- RxJS & Signals
- D3.js
- Spring Boot
- PostgreSQL + pgvector
- Node.js & Express
- Web Workers & WASM
- Server-Sent Events
- Docker
- Ollama & Qwen
- Open Source
- R&D
- Technical Documentation
Mar 2022 — Jul 2022
Internship — Full time
Full Stack Developer · NTT DATA Europe & Latam
As a Full Stack Developer at NTT DATA Valencia, I contributed to the end-to-end development of a corporate personnel management solution. I took part in implementing REST services on the backend and developed an interactive frontend module focused on dynamic visualization of structured data, ensuring code maintainability under the team's quality standards.
- Angular
- TypeScript
- Java
- Spring Boot
- Scrum
Mar 2018 — Jun 2018
Internship — Full time
Data Analyst · Sothis
Database analysis and design for the compensation plan of the UNE of Consulting and Information Systems (CSI) and the UNE of Automation and Control (AC).
- MySQL
- Power BI
Projects
Candle AI
Professional-grade candlestick charting and technical analysis dashboard that runs entirely in the browser — no accounts, no subscriptions, no data leaving your machine. Combines a deterministic Bayesian confluence engine (scoring 15+ candlestick and chart patterns across 7 signal hierarchies) with an optional AI narrative layer that explains the analysis in natural language. The LLM never overrides the quantitative engine — it narrates the results. Built with Angular 22 (Zoneless + Signals), Lightweight Charts v5, Web Workers for non-blocking computation, IndexedDB caching, Pine Script v5 export, and 160 Vitest + 5 Playwright E2E tests. Developed using prompt-forge agentic infrastructure across 9 epics as a solo developer.
- Angular 22 (Zoneless + Signals)
- Bayesian Confluence Engine
- Web Workers (Indicators)
- Lightweight Charts v5 (TradingView)
- LLM Integration (Ollama/DeepSeek/OpenAI)
- Pine Script v5 Export
- Playwright E2E + Vitest (165 tests)
prompt-forge
Agentic infrastructure toolkit that gives AI coding agents persistent memory across sessions. Designed to solve a real problem: AI agents are stateless — they forget lessons learned, repeat mistakes, and waste tokens rediscovering the same solutions. prompt-forge captures recurring errors, API changes, and workarounds as structured issues that are automatically promoted to permanent skills after 3+ occurrences. Zero runtime dependencies, no build step, no backend — just Markdown files that any AI coding agent (Copilot, Claude Code, Codex) can read. Multi-provider support with progressive disclosure (~700 tokens overhead across 7 skills). Built Candle AI on top of it — the infrastructure that made 9 epics of solo development manageable with consistent quality and knowledge retention.
- Agentic Infrastructure
- Agent Skills (agentskills.io)
- Knowledge Management
- Developer Tooling
- Git-native Persistence
- Progressive Disclosure
- Multi-provider
TestMaker
Secure exam platform with multi-user access and encrypted question banks (AES-256-CBC). Built with Angular 21 (Signals) and a Vercel serverless API. Features authentication, topic-based session configuration, advanced keyboard shortcuts, professional PDF export of results, ES/EN internationalization, and 18 Vitest unit tests. Built with AI assistance to solve a real need: studying with questions protected against unauthorized sharing.
- Angular 21 (Signals)
- AES-256-CBC
- Tailwind CSS
- Serverless API
- PDF (jsPDF)
- Vitest (18 tests)
- i18n ES/EN
Token Cost Analyzer
Client-side LLM prompt cost and tokenization analyzer. Runs tiktoken WebAssembly inside a Web Worker for non-blocking tokenization. Visualizes token distribution with D3.js bar charts and provides cost estimates for 10 LLM models. Built with Angular 19.1.8 (zoneless change detection), TypeScript 5.7.3, and D3.js 7.9.0. Published as open source in 2026, preserving the original 2025 tech stack.
- Angular 19 (Zoneless)
- WebAssembly (tiktoken)
- Web Workers
- D3.js
- TypeScript
- RxJS & Signals
Cognitive Agent Orchestrator
Cognitive agent orchestrator implementing the ReAct (Reasoning + Acting) pattern with real-time streaming via Server-Sent Events. Features a deterministic state machine that reasons, invokes tools (calculator, weather, search), observes results, and streams every thought to an Angular 18 zoneless frontend. Dual LLM provider support: simulated mode and Ollama + Qwen 2.5 Coder 7B. Backend: Node.js/Express with TypeScript 5.5.4. Published as open source in 2026, preserving the original 2024-2025 stack.
- Angular 18 (Zoneless)
- TypeScript
- Node.js & Express
- Server-Sent Events
- ReAct Pattern
- Ollama & Qwen 2.5
- Docker
Semantic Search Engine
Fullstack semantic search engine with vector storage in PostgreSQL via the pgvector extension. Implements cosine similarity search with HNSW indexing over 384-dimensional embeddings. Backend built with Java 21 and Spring Boot 3.2.6; frontend with Angular 17.3.12 featuring deferred result loading via @defer. Infrastructure orchestrated with Docker Compose. Published as open source in 2026, preserving the original 2024 stack.
- Java 21 & Spring Boot 3.2
- PostgreSQL + pgvector
- Angular 17 (@defer)
- TypeScript
- HNSW Indexing
- Docker Compose
Attention Matrix Explorer
Interactive visualization of transformer attention matrices (Query, Key, Value) inspired by the paper "Attention Is All You Need". Renders real-time heatmaps with D3.js (Viridis color scale, tooltips) and offloads matrix computations (softmax, transpose, multiplication) to a dedicated Web Worker. Built with Angular 16.2.12, Signals (Developer Preview), and D3.js 7.8.5. Published as open source in 2026, preserving the original 2023 stack.
- Angular 16 (Signals Preview)
- D3.js (Heatmaps)
- Web Workers
- Transformers (Q/K/V)
- TypeScript
My Portfolio
Personal portfolio built with Angular 19.1.7 and Tailwind CSS 4.1, featuring SSR, i18n (ES/EN), dark/light mode, and lazy-loading for a fast, accessible experience. Built with Angular Universal and Express.js for Server-Side Rendering in production.
- Angular 19.1.7
- Angular Universal (SSR)
- Tailwind CSS 4.1
- ngx-translate
- Express.js
- PostCSS
Palaze (BSc capstone · legacy)
Native web component library built with Stencil.js. Lightweight, easy-to-implement components with fast loading. Installable via npm or CDN. Includes dedicated documentation. Created as a final degree project to dive deep into web development.
- Stencil.js
- Web Components
- TypeScript
- HTML & CSS
- Figma