ML Engineering · Learning · Madrid
Learning machine learning engineering in Madrid
I am an AI and machine learning student in Madrid, and my goal is to become an ML engineer. While I get there, I practise as if I already were one: I build data pipelines, train and evaluate models, serve them behind an API and monitor them. From notebook to system, with the code public at every step.
Engineering, not demos
The difference between a notebook and a system is the engineering in between: leakage-free validation, honest evaluation, reproducible pipeline, API with input schemas, drift monitoring and a version registry. That is what I practise in every project.
Tabular, NLP and LLMs
Gradient boosting for tabular data, transformers and embeddings for text, and LLM applications when the problem calls for it. I implemented a transformer from scratch in PyTorch precisely to understand what is inside when something breaks.
Deployment as part of the design
I try to design every model thinking from the start about how it will be served: what input format, what decision threshold given the cost, how to version and how to detect when new data drifts away from training.
Learning in public
All my projects are on GitHub with a README, tests and their evaluation. I also publish a suite of CLI tools for ML workflows on PyPI. Learning in public forces a level of finish that a private notebook never demands.
Feedback, a question or an opportunity?
All my work is on GitHub and documented on this site. If you want to tell me something or think I could fit your team, write to me.
Write to meFAQ
What does a machine learning engineer do exactly?
They design and build the full system: collect and prepare data, define the problem correctly, train and evaluate the model, deploy it and monitor it. It is different from a pure data scientist (more exploratory analysis) or a pure data engineer (more data infrastructure). It is the profile I am studying towards.
What are you studying right now?
The mathematics behind ML with technical books (Mathematics for Machine Learning, Elements of Statistical Learning, PRML), and the engineering side by practising it: every concept I read, I try to land in a real project. Alongside, I contribute at Cylstat and to open source projects.
What technologies do you use?
Python with scikit-learn, XGBoost and LightGBM for classical ML; PyTorch for deep learning and NLP; FastAPI and Pydantic for APIs; Docker and GitHub Actions for infrastructure; Prometheus and Grafana for monitoring. For LLMs, the OpenAI and Anthropic APIs and HuggingFace models.
What can I see of your work?
Nine end-to-end projects at github.com/delcenjo (credit risk with full MLOps, a transformer from scratch, RAG, a SQL agent, LLM evaluation) and a suite of CLI tools at github.com/jmweb-org, published on PyPI with tests and CI. This site documents each one.
Are you looking for an internship or a junior role?
Yes. An internship or a first junior role in data or machine learning, in Madrid or remote. If you think someone who learns like this would fit your team, write to me through the contact form.