Where language expertise meets production AI
I'm Rubén de la Fuente, an AI and language technology consultant based in Madrid.
I help knowledge-work teams — in language services, content, training and professional services — adopt AI in real workflows. Not in pilots that die in PowerPoint.
I started in language work in 2001. Spent seven years building PayPal's Machine Translation program across every language they shipped in. Seven more as Senior Data Scientist on their platform engineering organization. The thread running through all of it: making language and data useful at scale.
AI training & enablement
Workshops and in-company programs for teams adopting AI
AI workflow audits
Where AI reduces friction in real processes — and where it can't yet
LLM & language technology evaluation
Evaluation frameworks, MT assessment, structured output quality
Data engineering & BI
Unstructured text and operational data into analytics-ready systems
Currently working with Cálamo & Cran and teaching AI and NLP at UDIMA.
Discovery layer for the Claude Skills ecosystem. Browse, compare, and check whether a skill already exists before building a new one — gap checker included.
LangGraph multi-agent pipeline that scores Spanish administrative documents against 41 plain-language criteria. Five evaluators running in parallel — results in seconds.
Chrome extension and standalone HTML tool that analyzes text against plain-language principles from the Government of Aragon Style Manual. Built for language professionals working with public administration documents.
Text-to-SQL analytics over operational data using a fully local LLM stack — Ollama, Vanna, DuckDB. Ask plain-English questions over DORA metrics; confidential data never leaves the machine. MSc thesis project.
What's Your Pick: RbMT, SMT or Hybrid?
Sampling for Machine Translation Evaluation
Postedició, canvi de paradigma?
From translator to data scientist
Engine-agnostic machine translation at PayPal
Most AI in knowledge-work dies in pilot.
The hard part isn't the model. It's the workflow.
Twenty-five years watching language and data problems get solved — and not solved — teaches you where the real friction lives. It's rarely in the technology.
I focus on making AI adoption structured, measurable, and maintainable — with people who actually use the output at the centre of the design.
I take on a small number of consulting, training and advisory engagements at a time.
If you're working on AI adoption for a knowledge-work team and think it might be a good fit, feel free to reach out.