Data scientist & AI engineer — building autonomous agents, computer-vision systems, and the data pipelines that turn raw data into decisions.


Full-stack AI SaaS that turns natural language into complex formulas (DAX, Excel, SQL, LaTeX) at 90%+ accuracy. Next.js 14, MongoDB Atlas, Claude API, JWT + Google OAuth, Stripe billing — 95+ Lighthouse on Vercel.

Production LangGraph agent doing end-to-end data analysis on CSV/Parquet from natural language. Understand→Plan→Execute→Validate→Synthesize state machine, self-healing loop (92% chart accuracy), and a 5-layer code-execution sandbox.

MSc thesis: a deep-learning model performing real-time object detection on live basketball video streams — part of the AI & Visual Computing programme.

At ELPEN Pharma: integrated and automated 150+ monthly reports into Power BI with Python pipelines for Excel processing, plus a redesigned performance-based bonus formula.
I turn data into decisions and products — from Power BI reporting at scale to autonomous AI agents and computer-vision models.
Currently CRM & Sales Data Scientist at ELPEN Pharma, where I automated 150+ monthly Power BI reports and built Python pipelines for data processing and analysis.
My background spans data analytics at IQVIA and Syscode, an MSc in Computer Science from the University of Piraeus, and an ongoing MSc in AI & Visual Computing (Univ. of West Attica & Univ. of Limoges).
I build in the open — an autonomous LangGraph data agent, and Querlytics, a full-stack AI SaaS that turns natural language into formulas.

Inspect the AI-generated 3D model from the homepage hero — orbit it, flip materials, and watch live render stats.

The GPU shader that powers the hero fallback, opened up — tweak grid density, wave physics, and colors in real time.