Biomedical Engineer (MSc, cum laude) | Medical Imaging AI | Research Software Engineering | Neurotechnology
I develop research-grade software and applied AI tools at the intersection of biomedical engineering, machine learning and real-world systems. My work spans medical-image analysis, EEG signal processing, reproducible experimental pipelines, and software/device interoperability.
I earned an MSc in Biomedical Engineering (110/110 cum laude) from the Università degli Studi “G. d’Annunzio” Chieti–Pescara. During an Erasmus+ research traineeship at g.tec medical engineering (Austria), I worked on EEG-based mental-workload assessment using N-back tasks.
Based in: Brescia, Italy
Interests: Biomedical R&D · Computer Vision · Neurotechnology · Applied AI · Scientific Software · Connected Systems
Archived release v0.1.7 — DOI: 10.5281/zenodo.23325290
Independently developed, single-author open-source Python research software. SpineLocus locates 11 thoracic vertebral landmarks (T1–T11) in digitally reconstructed radiographs (DRRs), combining YOLO-Pose with optional HRNet refinement.
Beyond model inference, the project implements subject-disjoint data partitioning, training configurations, full-image evaluation, ensemble methods, automated tests, and dataset/provenance audits.
Reported development-validation baseline: 82.31% mean subject PCK@5 and 6.424 px mean localization error, evaluated on 750 images from 15 subjects. This validation set was used during method development; these figures are not independent held-out test results.
Source code & documentation · Versioned archive / DOI · AGPL-3.0-only · v0.1.7
Research software only; not validated for clinical use. Imaging data and trained model weights are not distributed.
| Project | Engineering focus |
|---|---|
| EEG Mental Workload Analysis | MSc EEG research: N-back paradigms, ERP analysis, signal processing and mental-workload classification. Reproducibility-oriented research repository; distinct from SpineLocus. |
| Glucose–Insulin Modelling & Control | Biomedical system simulation with MATLAB/Simulink, PID and fuzzy control. Academic modelling project; not for clinical use. |
| MAGI / BioAudit | Experimental multi-model AI orchestration and biomedical research-auditing methodology. Research prototype. |
| Sanag Control for iOS | Original SwiftUI/CoreBluetooth application: reverse-engineered BLE interoperability, EQ presets and iOS Shortcuts. Hardware-tested on Sanag B21S. |
| UplinkWitness | Self-hosted Linux/network diagnostics with multi-layer connectivity evidence, dashboard and Raspberry Pi validation. |
| Echo LiveView Bridge | Experimental WebRTC/H.264 device-interoperability and video-stream research. Proof of concept. |
- Medical imaging & AI: Computer Vision, landmark localization, YOLO-Pose, HRNet, PyTorch, reproducible model evaluation.
- Biomedical research: EEG, ERP analysis, biosignal processing, experimental protocols and biomedical modelling.
- Scientific software: Python, scientific computing, scikit-learn, dataset integrity, testing, Git and CI.
- Systems & development: Swift/SwiftUI, CoreBluetooth, Linux, Raspberry Pi, network protocols and interoperability.
- Modelling & control: MATLAB/Simulink, dynamical systems and feedback control.
I'm interested in biomedical R&D, AI/computer-vision engineering and research software development, particularly where rigorous experiments need to become robust, maintainable software.
Projects vary in maturity; each repository documents its evidence, scope and limitations. Third-party contributions and methods retain their own attribution.