Design and develop algorithms that analyze athletes’ breathing patterns from earbud-microphone audio to optimize running performance. Apply signal processing and deep learning (including audio embeddings and transformer-based models) to turn raw audio into actionable performance insights, using TensorFlow, Keras, and PyTorch.

Develop statistical and AI models for financial time-series analysis, generating real-time insights to support trading and investment decisions. Build, train, and evaluate trading algorithms and deploy them into scalable production systems using Python and modern ML tooling.

Built machine learning and deep learning models for social-media trend analysis and network-traffic analysis, including event detection on Twitter and traffic-based user and mobile-device identification for platform security and profiling.

Led development of machine learning models for cybersecurity risk assessment in healthcare IT, translating advanced AI techniques into practical tools for identifying threats and vulnerabilities and enabling real-time, data-driven risk management.

Contributed to EU research projects (SEMIoTICS, CyberSANE, AI4HEALTHSEC, MARVEL) spanning IoT security, critical-infrastructure cybersecurity, AI for healthcare security, and multimodal analytics for smart cities. Developed novel detection and forecasting algorithms, applied large language models and explainable ML, and supervised bachelor’s and doctoral students.

Led acoustic source localization research in collaboration with Google, designing compressive-sensing algorithms for wideband localization and separation. This work led to an invited lecture at Google’s Mountain View HQ and a Google patent. Also developed a graph-theory-based room-geometry estimation method for a Huawei distributed-audio project, presented at Huawei’s Munich research center.

Developed signal processing algorithms for distributed sensor systems (DISFER, emergency response), invented sparse signal modeling methods for speech processing, and built multi-channel audio encoding and transmission algorithms.

Thesis: “Sparse and low-rank techniques for robust speaker recognition and missing-feature reconstruction”. Supervisor: Assoc. Prof. Athanasios Mouchtaris. GPA: 9.33/10. Defense slides PhD thesis

GPA: 7.33/10.

Thesis: “Multichannel audio modeling and coding for immersive audio based on the sinusoidal model”. Supervisor: Prof. Panagiotis Tsakalides. GPA: 9.71/10. First in class. MSc thesis

Diploma thesis: “Musical genre classification using statistical processing methods”. GPA: 8.00/10.