I build and ship production machine-learning systems — most recently applied-AI integrations at Tesla. Finishing my BSc in Artificial Intelligence at VU Amsterdam with a thesis on Spectral LoRA. ● Available July 2026.
Where I've shipped ML.
Tesla
Built and shipped an end-to-end NLP service that maps free-form customer complaints to vehicle diagnostics, deployed to Tesla's internal Kubernetes cluster. Owned the full stack: PostgreSQL schema, fine-tuned sentence-transformer embeddings, containerised FastAPI inference, and CI through Jenkins.
SkillLab.io
Kept a production BERT-based recommender (cosine-similarity matching over a user-skills database) reliable and current — diagnosed and fixed pipeline bugs, upgraded dependencies, and shipped improvements to the live service.
EY — Ernst & Young
Built an internal framework for benchmarking large language models on domain-specific tasks, with pluggable metric and dataset abstractions so new evaluations could be added without touching core code.
Martech Tribe
Shipped proprietary vendor-matching models (SVMs, BERT, and deep neural networks) that improved operational efficiency for enterprise clients.
Selected work.
Spectral LoRA
A novel fine-tuning method that allocates LoRA rank per transformer layer based on each layer's intrinsic dimensionality — measured via SVD stable rank. The goal: more parameter-efficient fine-tuning, and a clearer picture of how transformers represent information.
Deep-Q Learning for Schnapsen
A custom DQN agent in PyTorch with a self-play environment, experience replay, and Huber-loss optimisation for discrete action spaces.
Blue Origin · New Shepard Capsule
Designed and built a sealed capsule that ran viscosity measurements in microgravity on Blue Origin's New Shepard (launched May & December 2019). Co-authored the resulting scientific paper.
International Robot Olympiad — Gold
Coached a team to Olympic gold — technical design, programming, and problem-solving that ultimately led to a patented innovation.
Boeing Summer Coding School
Designed and ran individualised lesson plans alongside Boeing programme staff, tracking student progress and outcomes through the summer.
Youth Science & Technology Expo — 1st ×3
Built an LSTM for electricity-price prediction plus a news scraper to flag chaotic price events. Won 1st place in all three categories, including the People's Choice and Scientific awards.
What I work with.
Machine learning
Languages
Where I trained.
Vrije Universiteit Amsterdam
Coursework — click to expand
A bit about me.
I'm an AI developer & Machine Learning engineer finishing my BSc in Artificial Intelligence at Vrije Universiteit Amsterdam.
Since 2022 my work has shipped at Tesla, SkillLab.io, EY, and Martech Tribe — NLP services, transformer fine-tuning, and recommender systems running in production.
Outside the lab: Forbes 30 Under 30 Greece, patent holder for an assistive cane, 7th in the world at the World Robot Olympiad, and a contributor to Blue Origin microgravity launches.
Outside the lab.
Let's work together.
Open to research collaborations, ML engineering roles, and projects at the intersection of applied ML and theoretical insight. Available from July 2026.