Javier Matas
AI and Telecommunications Engineer
I am a Telecommunications Engineer, with deep passion for lifelong learning and a strong drive to craft innovative solutions for the future. I find joy in the fields of engineering, deep learning, and tackling new challenges. Furthermore, I possess valuable experience in translating abstract concepts into practical solutions through theoretical analysis and real-world problem-solving.
Additionally, I have a strong entrepreneurial and product vocation, with an interest in creating ML or DL projects that are a business success.
Technical Skills: Python, SQL, LangChain, Azure DevOps, FastAPI, Pytorch, NumPy and Pandas
Education
Master’s Degree in Telecommunication Engineering | Universidad Politécnica de Madrid
Specialised in Machine Learning, Deep Learning and Multimedia Data Science. Outside Spain it is called Electrical Engineering.
Bachelor’s Degree in Telecommunication Engineering | Universidad Politécnica de Madrid
Work Experience
Artificial Intelligence Engineer @ DXC Technology (august 2023 - Present)
I am currently collaborating as a team member involved in the development and implementation of Generative AI solutions.
- Development, implementation and performance tracking of a text processing tool, based on Azure Form Recognizer, that improves document table extraction and recognition by 15% in RAG solutions
- Collaboration in the development of a platform to accelerate generative AI solutions partnering with Ferrovial. Involved in the design and programming of libraries, the API using FastAPI, the Postgres SQL database, and the execution of MLOps tasks
- deation, creation and implementation of an interactive tool for the generation of executive summaries from internal documentation, using foundational LLMs and RAG. Used internally, it has reduced the time to create executive summaries by 50% to 70%
- Design and production of a tool that allows generating a battery of suggested questions from documents in a database. Improves the user experience with chatbots since it offers a variety of questions that can always be answered (by the chatbot) when they arise from the document database
Data Analist intern @ SDG Group (july 2022 - may 2023) During my last Master’s year I worked and studied part time. In SDG I provided technical assistance in the development of Business Analytics applications as well as I performed Data Extraction, Transformation and Loading tasks.
Personal Projects
Artificial Intelligence in financial markets. Master’s thesis
I have developed and implemented an artificial intelligence in a financial market simulation environment, which buys and sells shares automatically through a real broker. It is Based on Deep Reinforcement Learning, taking inspiration from AlphaZero and AlphaTensor. It has traded automatically on the Nasdaq, and has been awarded 3rd place in an international competition for quantitative trading algorithms, named Robotrader. Additionally, I had the honour to present it at the Madrid Stock Exchange, in a talk I gave on AI for trading (in Spanish). Starting at 1:32:45, at this link:
DQNTicTacToe
DQNTicTacToe is a full-stack web application where you can challenge a self-taught Deep Q-Network agent in a game of Tic Tac Toe. This project blends reinforcement learning with a user-friendly web interface, showcasing the potential of self-learning AI through interactive gameplay.
The DQN agent has learned the basics of Tic Tac Toe via self-play, but it continues to evolve with every game played against real users. By combining my reinforcement learning expertise with full application deployment, this project highlights both technical depth and practical implementation.
Tech Stack: Python, Flask (for the web interface), Gymnasium (for game mechanics), and PyTorch (for deep learning).
Live Demo: Play against the AI here
Open Source: Explore the GitHub repository
I welcome feedback, contributions, and, of course, your best moves again