Solutions Architect · Amsterdam
Building agentic AI systems and cloud infrastructure that turns complexity into clarity — at the intersection of technology and real-world impact.
About
My journey began with an endless curiosity for how things work — from computers and film cameras to ATM machines, always drawn to unravelling the underlying logic of technology. I thrive on creating robust solutions and tackling technical challenges, constantly seeking to learn new domains and master unfamiliar territory.
Originally from Greece, I embraced my love for real-world applications by co-founding a startup, an experience that fuelled my passion not just for technology but for the art of building businesses. To deepen my understanding, I pursued a master's degree in finance alongside computer engineering — an education that equipped me to bridge the worlds of technology and entrepreneurship with clarity and purpose.
Today, I call Amsterdam home and work at KPN, where I design agentic AI systems and cloud infrastructure for telecommunications — exploring the boundless possibilities that modern AI and cloud unlock for mission-critical operations. I devote my energy to tackling impactful, complex problems, collaborating and learning from others; each experience helping me uncover the influences and ambitions that define my journey.
Selected Work
Telecom · Cloud Infrastructure · Initiative
End-to-end architecture for a multi-tenant AI platform serving network operations: LLM gateway, vector database pipelines, semantic search, and knowledge management systems. Governs model evaluation, prompt engineering standards, and deployment strategy across teams.
AWS · Machine Learning
Co-authored AWS blog post and implementation guide for training a 6-billion-parameter GPT-J model using Amazon SageMaker's distributed model parallel library. Covers tensor parallelism, pipeline parallelism, and FP16 training to reduce cost and time at scale.
Read on AWS Blog ↗Open Source · Python
Hybrid ETF portfolio optimisation framework combining CVaR constraints, DCC-GARCH modelling, and Ledoit-Wolf shrinkage for enhanced tail risk management. Integrates VixFix-based regime detection, Monte Carlo scenario generation, and bootstrap backtesting to deliver superior drawdown protection during market stress events.
View on GitHub ↗Open Source · AWS
Serverless pipeline that uses AWS Step Functions to orchestrate Lambda functions calling Amazon Rekognition for face detection, then applies OpenCV to blur detected faces frame-by-frame in video — enabling privacy enforcement at scale. Deployed via AWS CDK with containerised Lambda runtimes.
View on GitHub ↗Expertise
Cloud Architecture
Agentic AI Systems
Finance & Quantitative Methods
DevOps & Delivery
Knowledge Management
Background
Contact
I'm always open to conversations about agentic AI, cloud architecture, or interesting problems at the edge of technology and business. Reach out through any of the channels below.
Whether you're working on a complex AI architecture challenge, curious about telco domain applications of agentic systems, or simply want to exchange ideas — I'd love to hear from you.
Particularly interested in connecting with engineers, architects, and researchers pushing the boundaries of what AI systems can do in production.
Say hello