Skip to content

About

Ten years of production systems, usually as the only architect.

I have spent ten years on production systems for government, healthcare and energy. On most of them I was the only architect, so I handled the database schema, ingestion, models, servers, firewall rules and go-live myself.

Often that was simply because nobody else was there to do it. The figures below come from systems that are still running.

Odysseas Tsakai
Odysseas Tsakai Ioannina, Greece

Some specifics

1B+ clinical records
National public-health analytics on a medallion architecture, with models and causal inference on top.
100M+ public documents
Ingested from Διαύγεια, ΓΕΜΗ, the Bank of Greece and ΕΣΗΔΗΣ. Multi-threaded PDF extraction, 12× faster than the sequential design.
20,000+ patients
The complete medical infrastructure for a mental-health service covering a whole city. Four years, sole technical resource.
400GB production database
Query restructuring: 90% faster under load. A Redis layer: another 50% off database pressure.

How I work

I stay through go-live

I do not hand over a diagram and leave. I write the schema, the pipeline, the model and the interface, and I stay until it runs in production. If it breaks later, I know where to look because I wrote it.

I work with teams and vendors

I have led small engineering teams, worked in cross-functional delivery alongside SAP and IBM, presented technical solutions to executive stakeholders, and managed vendors when their documentation stopped being accurate. Being the only architect meant owning the decisions, not working in isolation.

I find the cause before the fix

My PhD was in computer vision and CNN optimisation for medical imaging. I went into research because I wanted to understand why things work, not only that they do. I still debug that way: diagnostic queries first, fixes second.

I expect the data to be messy

In one national registry, two characters that look identical, one Greek and one Latin, had been silently corrupting the dataset for years. No architecture diagram shows that. It only shows up in production.

Education and research

PhD

Computer Science, University of Ioannina. Research in computer vision and CNN optimisation for medical imaging.

9.0/10

MSc in Computer Science and Networks, with distinction. BSc in Computer Engineering, 8.5/10, with honours.

10

Indexed publications. First author in Elsevier SoftwareX and IEEE BIBM. Publications

Experience

  1. 2026

    Current

    Principal Engineer, AI Operations & Innovation Hub

    Real Consulting

    Full-stack medical application with SAP integration, custom LLM and RAG pipelines, inside a cross-functional engineering team.

  2. 2025

    Current

    Architect & Senior Data Analyst

    D2A M.I.K.E.

    ETL pipelines and data platforms processing 5M+ records daily from European and national statistical sources into PostgreSQL, serving ministries and enterprise BI.

  3. 2025-Jul 2026

    AI Partner & Technical Advisor

    Vidavo

    AI strategy across healthcare and government: RAG pipelines for Greek ICD-10 medical coding, LLM integration, and architecture for Azure-based public-sector data infrastructure.

  4. 2016-2026

    Senior Principal Engineer, Technical Lead

    Q Base R&D

    Sole architect across national-scale production systems in healthcare, government, energy and blockchain. 100M+ documents, 20,000+ users, €2M+ in delivered project value. Built end to end: schema, ingestion, model, and the thing on the analyst's screen.

  5. Jan 2024-Jun 2024

    Senior Software Engineer

    ALXO I.K.E.

    Six-month technical transformation of a legacy digital services platform: backend refactoring, CI/CD modernisation, and engineering practice.

If you need a system like these designed or built, send me a message.

Send me a message