Pablo Olivares Martínez
ML Engineer
At Santander I help build the platform that takes teams from running experiments to shipping AI and agents into production. Sometimes I build models that go brrr.
Summary
ML Engineer selected for the Santander Data Science Talent Program. Proven ability to apply deep learning and achieve impactful results in data projects, along with a strong foundation in Mathematics and Computer Science.
Experience
ML Engineer @ Santander Bank
Selected for the Santander Data Science Talent Program, now contributing to advanced analytics and scalable data solutions within the Data & AI Models team.
- Developed an LLM explainability solution for predictive models by identifying key features from underlying data to justify commercial opportunities.
- Designed and implemented scalable data pipelines using PySpark for efficient data ingestion for model training.
- Developed a health insurance subscription prediction model achieving 92% AUC, leading to a 23% increase in customer acquisition.
- Designed and implemented a data quality control system for a Customer 360 project, improving data integrity and process execution time by up to 97%.
Education
Master of Science in Big Data & Business Analytics @ Escuela de Organización Industrial (EOI), Spain
Bachelor of Science in Mathematics @ University of Granada, Spain
Bachelor of Science in Computer Science @ University of Granada, Spain
Erasmus+ Programme @ University of Łódź, Poland
Projects
Topological Data Analysis in CNNs
Explored the integration of Topological Data Analysis (TDA) with convolutional neural networks (CNNs) to enhance understanding of CNN data manipulation, resulting in improved classification accuracy and generalization.
- Applied persistent homology techniques to analyze data structure during CNN processing.
- Proposed topological regularization in models like EfficientNet-B0 and DenseNet-121.
- Awarded 'Best Bachelor Thesis 2024 Promotion' for outstanding work.
Semantic Segmentation for Urban Mobility
Developed a deep learning solution for parking space detection in the city of Granada using semantic segmentation techniques, contributing to urban planning and mobility improvement.
- Created an image segmentation dataset with a novel data augmentation technique tailored for parking detection.
- Implemented architectures like PSPNet and DeepLabV3+, achieving an 80% F1-score in validation.
Skills
AI & ML
Tech Stack
