CV
Basics
| Name | Georgios (George) Kontogiannis |
| Professional Title | PhD Candidate |
| g [dot] kontogiannis [at] ac [dot] upatras [dot] gr |
I develop machine learning algorithms for respiratory health diagnostics, with a focus on audio modeling, edge deployment, and robustness across data modalities.
Experience
-
2026 - present Patras, Greece
ML Researcher
CTI Diophantus — EN (POLLA) PLO
- Research contract developing ML/DL systems to optimize Greece’s nationwide school textbook distribution ecosystem under the new multiple-textbook policy.
- Predictive models for student population, textbook demand, and inventory forecasting; school clustering by textbook selection patterns; simulation of logistics and what-if supply-chain scenarios; integration of the developed tools into the Publications Directorate IT infrastructure.
- Scientific Coordinator: Dr. Christos Manolopoulos
-
2022 - present Patras, Greece
Research Scientist
EU Horizon SynAir-G (Grant No. 101057271)
- Disrupting Noxious Synergies of Indoor Air Pollutants and their Impact in Childhood Health and Wellbeing, using Advanced Intelligent Multisensing and Green Interventions (Grant Agreement No. 101057271).
- Investigation of synergistic interactions between indoor air quality metrics and physiological signals from wearable devices using multivariate and temporal causal models.
- Scientific Coordinator: Prof. Nikos Papadopoulos
-
2024 - 2025 Thessaloniki, Greece
Research Scientist
Pfizer CDI — EyeAI Research Project
- Supervised dataset curation, annotation workflows, and quality control for ocular imaging.
- Developed an attention-gated UNet++ architecture for automatic sclera segmentation from smartphone-captured eye images.
- Quantized and deployed the model on-device for real-time smartphone inference (TensorFlow Lite / ONNX).
-
2023 - 2024 Thessaloniki, Greece
Research Scientist
Pfizer CDI — Voice-Based Diagnostics (VBD-2)
- Designed and deployed an end-to-end, on-device AI pipeline for continuous, passive respiratory screening with real-time cough detection and disease analysis.
- Processed 10-second tracheal and chest-wall auscultation recordings from the built-in microphone, with percussive noise removal and frequency-spectrum correction to mitigate domain shift across auscultation sites and microphone devices, for asthma and COPD detection.
- Developed a ResNet-34 classifier with a probabilistic voting mechanism for robust, explainable disease identification, plus deep generative models for cough and lung-sound synthesis to address training-data scarcity.
-
2022 - 2023 Thessaloniki, Greece
Research Scientist
Pfizer CDI — Voice-Based Diagnostics (VBD-1)
- Cough detection algorithms and abnormal/normal lung sound classifiers deployed for real-time smartphone inference using the built-in microphone.
-
2025 - present Rio, Greece
Research Scientist
University General Hospital of Patras — Dept. of Internal Medicine
- Supervised dataset curation, annotation workflows, and quality control for respiratory infection data (audio and imaging).
- Multimodal deep-learning research combining paired lung sound recordings, chest X-rays, and inflammatory biomarkers to distinguish bacterial vs. viral pneumonia and identify clinically meaningful audio biomarkers.
- Development of mobile and web-based clinical data acquisition tools for in-hospital deployment.
- Clinical collaborator: Prof. Dr. Dimitris Velissaris
-
2024 - present Rio, Greece
Research Scientist (Ocular Biomarkers)
University General Hospital of Patras — Dept. of Internal Medicine
- Supervised curation, annotation, and quality control of ocular images for jaundice assessment.
- Statistical and causal modeling of digital biomarkers and hematological indicators in patients diagnosed with jaundice.
- Study conducted under institutional ethics committee approval (University General Hospital of Patras IRB).
- Clinical collaborator: Prof. Dr. Dimitris Velissaris
Interests
Audio AI & Respiratory Sound Analysis: Digital signal processing, audio content analysis, deep learning, cough and lung sound analysis
Deep Generative Modeling: Synthesizing clinically meaningful medical audio, data augmentation and robustness
Robust & Deployment-Aware ML: Domain shift and dataset heterogeneity, edge deployment, trustworthy ML in real-world healthcare systems
Graph-Based Information Retrieval: Graph-based retrieval models enhanced by machine and deep learning
Education
-
2023 - present Rio, Greece
PhD Candidate
University of Patras
Computer Engineering & Informatics
- Advisor: Prof. Sotiris Nikoletseas
- Focus: audio AI for respiratory/cough sound analysis, multimodal AI modeling, deep generative modeling for medical audio, robust and patient-centric diagnostic pipelines for edge-deployed clinical applications
-
2018 - 2023 Rio, Greece
MSci (Integrated Master)
University of Patras
Computer Engineering & Informatics
- Advisor: Prof. Christos Makris
- Thesis: A Study on Improvement of Graph-based Information Retrieval Models through Machine and Deep Learning Techniques
- Thesis research was extended and published as a peer-reviewed journal article in Expert Systems with Applications (ESWA).
Teaching
- Guest Lecturer — AI Algorithms and Applications for IoT (Spring 2026). University of Patras, Dept. of Computer Engineering & Informatics; course coordinator Prof. Sotiris Nikoletseas. Designed and delivered a six-lecture applied ML module for IoT time-series data, developing all materials from scratch: LaTeX slide decks, Python demos, and a graded end-to-end project on the PAMAP2 activity recognition dataset.
- Lab Instructor — Analog Electronics Laboratory (Sep 2021 – Jan 2022). University of Patras; supervisor Dr. Georgios Oikonomou. Guided undergraduates through hands-on analog circuit experiments, theory sessions, and result analysis.
- Lab Instructor — Digital Electronics Laboratory (Feb – Jun 2022). University of Patras; supervisor Dr. Georgios Oikonomou. Guided undergraduates through hands-on digital circuit experiments, theory sessions, and result analysis.
Talks & Presentations
- EUSIPCO 2026 — “Metadata-Conditioned Audio Transformers for Adaptive Respiratory Sound Classification.” G. Kontogiannis, P. Tzamalis, S. Nikoletseas.
- DCOSS-IoT 2024 — “Exploring the Impact of Synthetic Data on Human Gesture Recognition Tasks Using GANs.” G. Kontogiannis, P. Tzamalis, S. Nikoletseas.
Awards
-
2023 University of Patras PhD Research Scholarship
University of Patras
Competitive scholarship awarded for PhD research in Computer Engineering & Informatics (2023–2025).
Skills
Machine Learning & Deep Learning: PyTorch, TensorFlow, Ajax, Scikit-Learn, Hugging Face Transformers, TensorFlow Lite, ONNX, MLflow, Weights & Biases
Signal Processing & Computer Vision: librosa, OpenCV, NumPy, SciPy, Pandas, Matplotlib, Seaborn
Programming & Scripting: Python, Java, C, C++, MATLAB, Bash
DevOps & Infrastructure: Docker, Docker Compose, Nginx, Systemd, Linux Server Administration
Mobile Development: Flutter, Android SDK
Databases & Storage: MySQL, PostgreSQL, Cassandra, MongoDB, Elasticsearch, MinIO
Development Tools: LaTeX, Git, GitHub, Atlassian Suite
Languages
Greek : Native
English : Proficient (Michigan ECPE)