Milagro Teruel

PhD in Computer Science
Data Scientist

Quick Bio

I'm a Computer Scientist based in Córdoba, Argentina. obtained my PhD in Computing Science at Universidad Nacional de Córdob in 2019, under the direction of Laura Alonso Alemany and Marcelo Errecalde.

As a computer scientist I do mostly Machine Learning, and even though I didn't wanted to be trapped by the Deep Learning hype, I ended up doing Deep Learning anyway. My thesis work was centered on neural embeddings and different representation learning techniques for educational data mining. After that, I specialized in modeling problems: defining specifications, obtaining and analyzing the data, planning and executing experiments, and wrapping up the product and results. My work is not just training a model, is delivering a comprehensive solution.

As a teacher... the faces of "ohhh that's how a recursive function works" are priceless. I try to always find time to teach, maybe even make a difference for someone. I strongly believe that education can change your life and add more value than any other thing to your personal capital.

On a completely unrelated topic, I love things that look pretty, and if I can combine that with programming I'm happy! Humans are visual, and if you want to effectively communicate something, start by making it appealing. That's why I like (and now teach) data visualization.

As a person, I'm in the nerverending story of doing as I wish. More often than not, the hard part is not the doing, is knowing what you wish. Things that I consider important are those who make our existance better, individually and socially, and those are quite hard questions to ask. In the meantime, I love huge puzzles, mandalas, origamis, and read as many books as I can find. That says everything you need to know. I sing too, for the detriment of everyone sharing the same space with me at the time.


Representation Learning for Educational Data Mining

This is my main PhD topic. We are studying automatic representations for low level data, extracted from logs of educational material such as MOOCs or tutoring systems. In particular, we modify the traditional recurrent architecture to inject domain knowledge into the model and represent students along with course elements in the same latent space. We look for results in interpretability to aid teachers and instructional designers to handle large volumes of data.

Read More

Argumentation Mining

Our aim is to automatically extract argumentative structures from legal text, in particular judgements of the European Court of Human Rights. We have been experimenting with deep learning models with attention to solve this complex problem using very small datasets. This is a work in collaboration with WIMMICS team, INRIA, in the French city of Sophia Antipolis.

Read More


Teaching is one of the most rewarding activities of my week. I started working as a Student Assistant in 2013, and since then I've tried to always find the time to do it. I'm currently holding an Assistant Teacher position at FaMAF. Some courses that I've been involved with are:

  • Algorithms and Data Structures I and II
  • Databases
  • Operative Systems
  • Programming Paradigms

Now I also teach at the Diplomatura en Ciencia de Datos, Aprendizaje Automático y sus Aplicaciones, an undergrad course on Data Science and Machine Learning. We have been developing this initiative since 2016 and it finally became true on 2018. Along with Soledad Palacios we teach Data Analysis and Visualization, and with Cristian Cardellino we teach Deep Learning.

Work Experience

During and after my PhD I've worked in several Data Science projects, both as part of a team and leading some. But I also have a strong computer science core underneath that: I worked as a Software Developer for several companies while I was doing my Master and PhD. Always for a couple of months, but they were all amazing experiences and I got to learn from the best programmers. Most of my industry experience is related to Web Development, what I like to call the "frontend of the backend". I also worked for some time analyzing data generated by mobile phones to detect relevant patterns.

The most important company I've worked for is Google, where I did three internships. I also worked in startups in Córdoba like Machinalis, and as volunteer for an open source project, Oppia.

Check linkedIn for updates.