CV
Contact Information
| Name | João Cardeira |
| Professional Title | AI Researcher & PhD Student |
| jaca.pereira@campus.fct.unl.pt |
Professional Summary
PhD student at NOVA LINCS working on large vision-language models, video and anomaly understanding, and multilingual LLMs. Core team of AMALIA, Portugal’s open LLM.
Experience
-
2024 - present Portugal
AI Researcher
AMALIA LLM Project
- Training and evaluation of the AMALIA-VL model
- Lead author of PorTEXTO, the first benchmark for modern pt-PT OCR evaluation
- Contributed to training pipeline, pt-PT multimodal data mix, and evaluation suite (amalia-vl-eval)
-
2025 - 2025 Covilhã, Portugal
Teacher — Data Science and Machine Learning
Universidade da Beira Interior
-
2020 - 2020 Lisbon, Portugal
Student Internship
Departamento de Informática, NOVA School of Science and Technology
- Development of a CRM strategy for DI@FCT
Education
-
2024 - present Lisbon, Portugal
PhD
NOVA School of Science and Technology, Universidade NOVA de Lisboa
Generative AI and Computer Science
- Research: large vision-language models, video anomaly understanding, multilingual LLMs
- Lab: NOVA LINCS
-
2017 - 2023 Lisbon, Portugal
Integrated Masters (BSc + MSc)
NOVA School of Science and Technology, Universidade NOVA de Lisboa
Computer Science and Engineering
- Thesis: Real-Time Human Action Localization in the Wild
-
2021 - 2021 Dresden, Germany
Exchange Program
Technische Universität Dresden
Computer Science and Engineering
Awards
-
2024 Best Poster Award
RECPAD 2024
Publications
-
2026 FineVAU: A Novel Human-Aligned Benchmark for Fine-Grained Video Anomaly Understanding
Proceedings of the AAAI Conference on Artificial Intelligence
-
2026 AMALIA-VL: A Native European Portuguese Open-Source Vision and Language Model
arXiv preprint arXiv:2606.19100
-
2025 Self-ReS: Self-Reflection in Large Vision-Language Models for Long Video Understanding
IEEE International Conference on Multimedia and Expo (ICME)
-
2026 PorTEXTO: A European Portuguese Benchmark for Visual Text Extraction
arXiv preprint arXiv:2606.19096
-
2025 Chain-of-Anomaly Thoughts with Large Vision-Language Models
arXiv preprint arXiv:2512.20417
-
2025 Towards Event-Driven Evaluation of Surveillance Video Understanding Using Natural Language
Iberian Conference on Pattern Recognition and Image Analysis (IbPRIA)
-
2024 Zero-Shot Action Recognition in Surveillance Videos
arXiv preprint arXiv:2410.21113
Skills
Machine Learning & Deep Learning: PyTorch, Transformers, LLMs, Vision-Language Models, Video Understanding
Data Science: Pandas, Matplotlib, Data Analysis
Programming: Python
Languages
Portuguese : Native
English : Full Professional
Spanish : Elementary
Interests
Research: Video Anomaly Understanding, Large Vision-Language Models, Multilingual LLMs, Open-Source AI