Panagiotis Michael

Data Scientist

panagiotis.michael133@gmail.com
orcid.org/0009-0008-9660-6117

Experience

ML Researcher – AdaptoFlow (EU Horizon Project)
University Of Nicosia, AI Lab

07/2024 – 05/2025

Contributed to adaptive AI/ML inference and energy-aware model swapping for geo-distributed edge environments.

Developed real-time data stream processing techniques to optimize resource usage while preserving service quality.
Achieved 68% reduction in inference delay and 27% energy savings, while maintaining accuracy within 1% of baseline models.

ML Engineer – University Of Nicosia, AI Lab

10/2023 – 01/2024

Designed, developed, and deployed an LLM-powered chatbot for Data Science MSc program inquiries.
Managed data acquisition, preprocessing, and integration into a conversational pipeline.

ML Engineer (Intern) – AC Goldman Solutions & Services Ltd

06/2023 – 08/2023

Curated and preprocessed legal corpora for NLP applications.
Fine-tuned LLMs to improve semantic search in legal document retrieval systems.

Data Scientist (ιδΕΚ Project) – University Of Nicosia

04/2023 – 06/2023

Built an end-to-end ML pipeline and deployed a REST API for hotel customer segmentation.
Achieved over 95% classification accuracy using open-source data.

Education

MSc. Data Science
University Of Cyprus
09/2024 – present

BSc. Data Science
University Of Nicosia
09/2020 – 06/2024

Publications

Towards Low-Cost and Energy-Aware Inference for EdgeAI Services via Model Swapping.
IEEE International Conference on Cloud Engineering (IC2E), 2024

Demetris Trihinas, Panagiotis Michael, Moysis Symeonides

Proposed ModelSwapper, a Python toolkit for dynamic model selection, achieving up to 68% latency and 27% energy reduction using EfficientNet and BERT.

Evaluating DL Model Scaling Trade-Offs During Inference via an Empirical Benchmark Analysis.
Future Internet, 16(12), 468, 2024