06/06/2026 updated
100 % available
Freelance Computational Engineer | ML, Simulations & Robotics
Nicosia, Cyprus
Worldwide
B.Sc. Physics, TUMAbout me
I'm a computational engineer. I build things at the intersection of physics, engineering, and software: machine learning pipelines, robotics control systems, numerical simulations. Wherever the bottleneck is turning a hard problem into working code.
RustData AnalysisC++ (Programming Language)Computational Science and EngineeringFinite Element MethodsPython (Programming Language)Machine LearningNumerical Computing
Machine Learning & Data Science
Expertise in building machine learning pipelines using PyTorch, scikit-learn, XGBoost, BERTopic, LDA, Hyperopt. Application of statistical models, graph auto-encoders, neural networks, transformers and NLP methods to real-world research problems.
Robotics & Embedded Systems
Development of autonomous robotic systems using ROS, RealSense, Hailo-8, YOLOv8, ByteTrack and OSNet. Implementation of real-time human following, obstacle avoidance, and re-identification on embedded hardware such as Raspberry Pi 5.
Numerical Simulations
Finite element simulations using FEniCSx and Gmsh for modeling physical phenomena such as heat diffusion and stress fields. Validation of simulation results against analytical solutions.
Programming Languages
Proficiency in Python, Rust, C++, Java, and JavaScript for a wide range of engineering and software development tasks.
Time-Series Analysis
Development of the open-source Python library reidfo for time-series regime identification, including feature engineering, statistical diagnostics, copula fitting, and jump model utilities.
Natural Language Processing
Application of LDA and BERTopic to large corpora of academic papers, including preprocessing, semantic cross-validation, and manuscript preparation.
Web & Tooling
Experience with Linux, Flask, Socket.IO, React, and Git for building and deploying software applications and services.
Languages
GermanGoodEnglishFluentGreekFluentRussianFluent
Project history
Developing an autonomous navigation system for reconnaissance in environments that are hazardous or inaccessible to humans. The robot uses LiDAR SLAM for localization and map generation, with a Visual Language Model as the high-level planner, enabling navigation via natural language commands.
Finite element simulation of laser-induced thermal spallation in rock using FEniCSx and Gmsh. Models heat diffusion and resulting stress fields. Validated against analytical solutions.
Modelled political risk as a regime-switching process. Built statistical jump models and graph auto-encoder approaches. XGBoost classifiers achieved out-of-sample recall 0.6-0.9 across regimes. Extended abstract accepted at LMDE 2025.