About Me

I am currently a data scientist at DareData where I focus on developing to accelerate and optimize automotive electrical cable design.

Previously, I was a visiting student at the Psychology Department and Neuroscience Institute at Carnegie Mellon University under Maggie Henderson, where I leveraged spatial frequency constraints derived from fMRI data to regularize computer vision models. Before that, I was a researcher at INESC-ID and Instituto Superior Técnico (IST) in the University of Lisbon, where I focused on developing a subcortical module for visual processing to improve model robustness to image perturbations while mirroring neuronal representations.

CV LinkedIn Scholar Github E-Mail

Work Experience

Data Scientist @ DareData
(Jan. 2026 - Present)

Building data pipelines and predictive models on historical lab data to accelerate automotive electrical cable design.

Visiting Student @ Carnegie Mellon University
(Nov. 2025 - Dec. 2025)

Using spatial frequency tuning profiles from fMRI data to regularize early representations of computer vision models.

Research Assistant @ INESC-ID
(June 2024 - Oct. 2025)

Improving standard backbone CNN architectures with biologically inspired front-end modules to improve brain alignemnt and robustness against image corruptions, distribution shifts and adversarial attacks.

Education

M.S., Computer Science and Engineering @ IST ULisboa, GPA: 19/20
(Oct. 2021 - Nov. 2024)

Specialization in Artificial Intelligence and Interaction & Visualization
Thesis: “Using Biological Features of Improve Deep Neural Network Models for Vision” (Grade: 20/20)

Erasmus Semester @ University of Amsterdam, GPA: 8/10
(Aug. 2022 - Jan. 2023)

B.S., Mechanical Engineering @ IST ULisboa, GPA: 16/20
(Sep. 2018 - Aug. 2021)

Projects

Explicitly Modeling Subcortical Vision with a Neuro-Inspired Front-End Improves CNN Robustness
Paper Code

Publications

Piper, L.; Oliveira, A. L.; and Marques, T. “Explicitly Modeling Subcortical Vision with a Neuro-Inspired Front-End Improves CNN Robustness” (2025). NeurIPS 2025 Link