CV

Profile

Bioscience Engineer with an additional Master’s in Artificial Intelligence. I work at the intersection of computer vision and sensor time-series modeling, currently building AI systems for structural damage detection (DINOv3) and multimodal Human Activity Recognition with IMU and radar sensors at Sentigrate. My broader experience spans medical imaging (ophthalmology, hyperspectral imaging), wearable biosignal processing, and deep learning on the edge, from foundation model fine-tuning to model compression and efficient inference. I bring strong cross-disciplinary skills, developed through research in both life sciences and AI.

Skills

Computer Vision: vision foundation models (DINOv3), semantic segmentation (SwinV2), object detection (YOLOv8), disease classification (Alzheimer, Glaucoma), video & hyperspectral imaging, feature extraction (texture), annotation platforms (Lirot.AI)

Time-Series & Signal Processing: multimodal HAR (IMU + radar), anomaly detection with foundation models (Chronos2), EEG/EMG processing, wearable sensor fusion

Edge AI & Model Efficiency: DL on the edge, model compression (quantization, pruning, knowledge distillation), efficient architecture design, ONNX export

ML/AI: CNNs, Vision Transformers (ViT), VLMs, self-supervised learning, domain adaptation, multimodal fusion, foundation model fine-tuning (HuggingFace)

Software Engineering: Python (PyTorch, OpenCV, NumPy), testing (pytest), CI/CD (Git/GitLab)

Communication & Languages: scientific writing & presentations, cross-disciplinary teamwork; Dutch (native), English (fluent)

Research Experience

Data Scientist @ Sentigrate – Nov 2025-present

SCOPR — AI-powered building inspection

• Facade crack detection and damage segmentation using DINOv3-based pipelines
• Exploring Vision Language Models (Gemma 4) for automated inspection report generation

FoG — Freeze of Gait detection for Parkinson’s patients (in collaboration with Epihunter)

• Wearable EEG-based algorithm to detect and predict motor freezing episodes
• On-device inference for real-time sensory cue triggering

DistriMuSe — Multimodal activity recognition for elderly care

• Sensor fusion across IMU wearables, radar, and localization beacons
• Real-time anomaly detection and activity classification for proactive care intervention; model compression for edge deployment

Traffic anomaly detection (in collaboration with Telraam)

• Anomaly detection in traffic time-series using Chronos2, a foundation model for time-series forecasting

Interdisciplinary research between Research Group Ophthalmology and Biomedical Signal Processing (KU Leuven) – Jan 2024-Sep 2025

Multimodal retinal imaging for early Alzheimer’s disease detection

• End-to-end pipeline (IQA, preprocessing, feature extraction, multimodal classification); best model achieved AUC 0.84
• Found novel texture biomarkers in hyperspectral retinal imaging
• Assessed repeatability and reproducibility of image biomarkers

Multi-Source Domain Adaptation for Glaucoma Classification

• Addressing domain shift across different imaging devices and demographics by leveraging Vision Foundation Models (VFM) pre-trained with self-supervised learning
• Using multi-source domain fine-tuning and implementing domain adaptation techniques

Artery and vein segmentation in fundus image and video data & subsequent parameter extraction

• Implemented custom loss functions to preserve correct topology, ensuring temporal consistency in video segmentation
• Assessed repeatability of extracted vascular parameters (PVBM toolbox)

Research Internship Clouds of Care, Gent – August – September 2023

• Research on the bi-directional link between sleep and dementia
• Developing 1D-CNN for sleep spindle detection in EEG sleep data

Education

• Master of Science: Artificial Intelligence, KU Leuven, 2022-2023
  • Thesis: Developing an automatic seizure detection algorithm to facilitate treatment of patients with epilepsy using multimodal wearable sensor data (EEG, EMG, ACC)
• Master in Bioscience Engineering: Human Health Engineering, KU Leuven, 2020-2022
  • Thesis: Developing a camera-based activity recognition algorithm by combining object detection (YOLOv8) with motion detection (GMM)
• Bachelor in Bioscience Engineering, KU Leuven, 2017-2020
  • Undergraduate Thesis: researched bacterial protein aggregates, their induction, localization, and impact on cellular function (Selected as a finalist for the Student Research Awards)

Publications

• Ghesquiere, M. et al., “Multimodal Retinal Imaging for the Detection of Early Alzheimer’s Disease: Combining Biochemical, Structural, and Vascular Biomarkers,” Bioengineering, vol. 13, no. 9, p. 1041, 2026, doi: 10.3390/bioengineering13091041.
• M. Ghesquiere et al., “Seeing the light: Hyperspectral retinal imaging as a surrogate marker for early stage Alzheimer’s disease,” Acta Ophthalmologica, vol. 103, no. S284, 2025, doi: 10.1111/aos.17369.
• Willems, S., Ghesquiere, M., Liu, D., Mounir, M., & Norton, T. (2022). ‘Development of a camera-based hybrid approach for the early detection of poultry red mite-related behaviors in individual laying hens’. Department of Biosystems, KU Leuven, Belgium

Conferences

• European Glaucoma Society (EGS) 16th congress – 1-4 June, 2024
• European Association for Vision and Eye Research (EVER) congress – 3-5 November, 2024

Volunteering

• Manager of the Faculty Bar, LBK, Leuven
  • Managing a team of 15+ people, communicating with external organizations, and solving any problems pertaining to the bar’s function
• Leader & Board Member, KLJ Hagelstein
  • Animator and active member of the KLJ youth movement organization