Welcome to my personal website 👋
ML Researcher and Professor
I explore how Artificial Intelligence can help
people/organizations make better decisions
Medical: Analysis of sequential data (EHRs, ECGs) and imaging (X-rays, fPET scans).
Industrial: Predictive maintenance for lithium-ion batteries, aircraft engines, jet fans...
Teaching: Assistant professor at the University of Oviedo. Gijón Campus.
Talks & Consulting: I help you identify how to integrate AI into your business.
Trusted by
Companies and organizations that I have work or collaborated with.
About Me – Nahuel Costa
Hello, my name is Nahuel (it means jaguar), I'm an ML researcher and professor at the University of Oviedo.
My research focuses on maximizing the potential of Machine Learning to anticipate possible outcomes and support decision-making in monitoring systems, particularly in scenarios with limited data. I work mainly on the biomedical field and in the industrial domain. I am particularly interested in developing models that are robust and effective, but also interpretable and easily accessible to non-ML/AI experts.
💬 I love sharing my work and insights on AI with a wider audience. If you’re interested in workshops, talks, or tailored training for your company or organization, feel free to reach out. You can explore the Training & consulting section for an overview of what I offer or contact me directly to discuss how we can create something that fits your needs.
Assistant Professor
University of Oviedo
Subjects I teach:
- Generative models
- Business Intelligence
- Data Visualization
- Algorithmics
Visiting Professor
University of Alberta
Visiting Professor
University of Alberta
Visiting Professor
CEU San Pablo University
Lecturer
University of Oviedo
-Business Intelligence
-Data Visualization
-Algorithmics
-Operating Systems
-Databases
-Programming methodology
-Introduction to programming
Visiting Researcher
Université de Montpellier
Visiting Researcher
University of Hawaii at Manoa
Research technician
University of Oviedo
Research intern
University of Oviedo
Research
Main topics and fields I work with
Industrial
- •Battery health monitoring and prognostics
- •Aircraft engine condition monitoring
- •Jet fan predictive maintenance
- •Physical systems modeling with digital twins
Medical
- •ECG analysis and cardiovascular disease prediction
- •Medical image analysis (X-ray, PET scans)
- •Electronic Health Records (EHR) analysis
- •Multimodal and Longitudinal patient data modeling
Others
- •Machine learning with limited data
- •Explainable AI methods
- •Chatbots and Retrieval-Augmented Generation (RAG)
- •Agent-based AI systems
- •Educational AI applications
Research Projects
Past and current research projects
Database audit for a future public health information system
FUO-24-202
Database audit of public health data based on the social determinants of health model.
Intelligent monitoring system for anomaly detection and operational optimization in Smartfan tunnel fans
FUO-25-065
Continuation of the tunnel fan monitoring project initiated in 2022 (FUO-231-22). In this phase, we are focusing on improving anomaly detection, not only independently but also in pairs of fans, as well as optimising fan operation.
Team member
ATLAS - GEOAI-Based Augmentation of Multi-source Urban GIS
NAC-ES-PUB-ASV-2025 PCI2025-163245
The multidisciplinary ATLAS project was proposed under the CHIST-ERA Call 2023. ATLAS aims to enhance the expressiveness and quality of Geographic Information Systems (GIS) by integrating data from multiple external sources of diverse nature and quality, with a case study focused on urban mapping and flooding. The project contributes to the broader effort of developing and applying tools and methods based on AI, ML and statistics in the water sciences.
Team member
Intelligent Computing For Disruptive Data
SEK-25-GRU-GIC-24-055
The use of incomplete, inaccurate or partial data in machine learning can lead to biased or inefficient decisions. Commonly, large data sets are used, selecting informative elements and discarding those that do not meet a minimum quality level. Our research group proposes an alternative approach for scenarios where data scarcity requires maximising every piece of available information, a frequent challenge in areas such as industry, finance, economics, and medicine.
Team member
Sustainable Computing in Limited Information Scenarios
MCINN-24-PID2023-146257OB-I00
This project is dedicated to developing new methods and applications to address ML challenges in environments characterized by sparse information. It gives priority to the use of low quality data, which is often overlooked. In alignment with Green AI principles, our approach includes designing methodologies that enhance computational efficiency and, as a result, reduce energy consumption. Furthermore, we are adapting these methodologies to address specific industrial challenges, focusing on sustainable practices such as extending the lifespan of equipment and efficiently managing resources like water and energy.
Team member
Observatory for the Implementation of the 2030 Agenda in the Spanish University System
MDSC-24-2024D179-Agenda2030
Working on CITIES DATALEX®, a software whose purpose is to improve access to legal regulations resulting from the application in actions in the urban environment and, in general, in the processes of sustainable urban and territorial development. Also, supervising the IT team in helping gather and process data for the main purpose of the project.
Team member
Training & consulting
For companies and institutions: AI consulting, courses, workshops and talks.
Private 1:1 consulting
1:1 guidance for companies/organizations that want to actually use AI, and stop getting lost in tools and promises.
Teaching
University teaching: study materials and projects
Generative Networks
View Materials →
Business Intelligence
View Materials →
Algorithmics
View Materials →
Innovation Projects
View Project ★
Mentored Students
Students I've guided in their research and academic projects
Internships
16 students at 10 companies
Samuel Camba Fernández
Prognosis of Degenerative Diseases Using Unsupervised and Partially Supervised Learning Techniques
PhD research focused on developing novel unsupervised and semi-supervised learning techniques for early prediction and prognosis of degenerative diseases.
Pablo Acinas Suárez
Development of a Conversational Assistant for the Automated Management of Small Businesses using Generative AI
Design and implementation of an intelligent voice agent capable of holding fluid, natural conversations in real time, with low-latency bidirectional audio and the ability to carry out complex tasks. The system was validated through a practical use case: managing restaurant bookings.
Alba Espinel Martín
ECG Analysis with Large Language Models: Study and Application of ICL
Study and application of in-context learning (ICL) to ECG interpretation tasks, with particular emphasis on assessing how well these models adapt to settings where specialised models fail due to data scarcity.
Jorge Valdenebro Álvarez
Study and Application of Domain Adaptation Techniques in Deep Learning Models
Analysis of various domain adaptation techniques for the detection of cardiovascular risk, where differences in patient characteristics and capture equipment can influence the accuracy of trained models. Evaluation of the effectiveness of the selected techniques to improve generalization between different electrocardiographic datasets.
Mario Rabanal Pérez del Río
Intelligent System for Access and Consultation of Legislative Documents
Web tool capable of providing clear, secure and relevant legislative information, assisted by AI. The tool includes the ability to process and understand both structured (sections, tables, regulations) and unstructured (free text, figures, reports) information, ensuring reliability and avoiding typical errors of interpretation or "alucinations" of current language models
Showing 1-6 of 13 students
Blog
Other projects, tutorials, and more.