Bálint Décsi
CTO at Proximata · also Builder at mesh.
Abstract
Engineer focused on the end-to-end data and ML systems lifecycle — from building cloud pipelines and lakehouse platforms to shipping machine-learning solutions with AI agents. Recently graduated from MS in Business Analytics at Central European University. Also love prototyping ML-driven products.
† experience‡ github§ linkedin
§1At a glance
- Current role
- CTO at Proximata (Vienna-based hacker lab)
- Also
- Builder at mesh.
- Education
- MS Business Analytics, CEU (2025–2026)
- Location
- Vienna, Austria / Budapest, Hungary
§2Interests & skills
[python][ai agents][ml engineering][apache airflow][apache iceberg][google cloud][deep learning][computer vision][geospatial analytics][xai][sql][data lakehouse][saas prototyping]
§3Work
The work I'd point to first — the venture I'm a co-founder of, plus case studies and demos I've shipped.
§3.1Live and clickable apps
Proximata
Co-founder & CTO — AI-native products lab
Vienna-based hacker lab building AI-native products. I co-founded Proximata and lead technical strategy as CTO — from infrastructure to venture-building and spinoffs.
[ai systems] [venture building] [infrastructure]
Open Drone Space
Open-source minded map for drone pilots — data pipeline engineer
An open-source minded map for drone pilots, showing Hungary's static drone airspace layers (no-drone zones, aerodrome zones, airspace). I engineered the data pipelines for different interval scraping/fetching from a dozen of sources.
opendrone.space ↗open-drone-space-data ↗
[data pipelines] [geospatial] [open source] [aviation]
Skills
Global skill library for my AI agents
The global skills I use for my agents across different projects — reusable, versioned skill definitions shared by every agent setup I run.
[ai agents] [open source]
§3.2Case studies
MSc capstone — Budapest rental prediction
ingatlan.com × CEU · end-to-end modelling report
Public-facing summary of my MSc Business Analytics capstone: temporal validation, geospatial enrichment (WorldPop + Sentinel-2 NDVI on H3), model leaderboard, SHAP diagnostics, and product recommendations.
[ml] [geospatial] [capstone]
Sovereign data platform
Data Engineer
Turning a 20-year-old telco data estate — full of sensitive customer data and bound by strict EU regulation — into a nimble, cloud-native lakehouse.
- Apache Iceberg lakehouse on Google Cloud with automated Airflow ingestion for the fixed-line segment.
- Scalable ML solutions and pipelines delivered through GitLab CI/CD.
[gcp] [bigquery] [python] [apache iceberg] [airflow] [gitlab ci/cd]
Domestic network optimization
Data Scientist
Greenfield geospatial machine learning for domestic retail-network optimization at Hungary's largest telco and banking companies. A small team covered end-to-end data and ML modelling, from feature store to client presentation.
- End-to-end delivery of ML models for external B2B clients, on a custom geospatial and location-based feature store.
- Unsupervised models for network-infrastructure optimization, with Explainable AI (XAI) to build business trust in model decisions.
- Automated web scraping for competitive intelligence and interactive geospatial visualizations for strategic network planning.
[python] [scikit-learn] [tensorflow] [gis] [bigquery]
Vodafone Hungary data migration
Data Engineer
Moving the data and analytics platform of Hungary's second-largest telco — ~3.8M residential and business subscribers after the UPC Hungary acquisition — to Google Cloud.
- Built ETL pipelines migrating critical on-premise data to cloud-based environments.
- Designed and implemented optimized data marts on a cloud-native lakehouse architecture.
- Worked with cross-functional stakeholders to define technical data requirements.
[gcp] [bigquery] [sql] [etl] [lakehouse]
§3.3Demos
Comics Factory
ML-driven SaaS for stylized comics with consistent characters
Generate stylized comics with consistent characters from a handful of reference images. Built at Hungary's first hacker space using Supabase, Firebase, and modern image-gen pipelines.
[llms] [image gen] [saas] [supabase]
CEU FeedForward
Anonymous student-feedback platform for CEU
MVP demo of a secure, GDPR-compliant student feedback platform — registration with a verified @student.ceu.edu email, structured ratings, free-text comments, and institutional dashboards.
[civic tech] [privacy] [higher ed]
PDF → Word
Convert a PDF to .docx, all in the browser
Drop a PDF in and download an editable .docx. Runs entirely client-side using pdf.js — no file ever leaves your machine.
[utility] [client-only] [pdf]
Unibridge
Onboarding companion for international students in Vienna
Multi-language prototype that walks new students through housing, health insurance, visa, and local admin. Built as a public-good civic-tech experiment.
[product] [i18n] [civic tech]
§4Experience
Roles, workplaces, and education — the full CV is downloadable as a PDF.
§4.1Employment
CTO
Proximata · May 2026 — Present · Vienna, Austria
- Leading technical strategy and execution at Proximata, a Vienna-based hacker lab.
- Driving product and engineering decisions from concept to production for AI-native products.
- Building tech solutions and spinoffs by enabling ambitious builders with infrastructure and resources.
[technical leadership][ai systems][product engineering][venture building]
Builder
mesh. · June 2025 — Present · Budapest, Hungary
- Member of Hungary's first hacker space.
- Engineering an ML-driven SaaS that generates stylized comics with consistent characters from images.
- Rapid prototyping using Supabase, Firebase, and modern full-stack tools.
- Leveraging AI agents (Cursor, Gemini CLI, Copilot) for accelerated, parallelized development.
[ai agents][ml prototyping][saas][supabase][firebase][llms]
Data Engineer
Deutsche Telekom IT Solutions HU · May 2024 — Jul 2026 · Budapest, Hungary
- Engineering scalable ML solutions and pipelines for the fixed-line business segment.
- Building a modern Data Lakehouse using Apache Iceberg on Google Cloud Platform.
- Developing automated Apache Airflow pipelines for robust data ingestion and processing.
[ml engineering][apache iceberg][gcp][airflow][data lakehouse]
Advanced Analytics Expert
Vodafone Hungary · Oct 2023 — Apr 2024 · Budapest, Hungary
- Optimizing network infrastructure using unsupervised machine learning techniques.
- Implementing Explainable AI (XAI) to drive business engagement and trust in model decisions.
- Developing interactive geospatial visualizations to support strategic network planning.
[unsupervised ml][xai][geospatial][python]
Discover Graduate Program
Vodafone Hungary · Sep 2021 — Oct 2023 · Budapest, Hungary
- Rotations: Data Engineer, Data Warehouse Specialist, Data Analyst, Data Scientist.
- Built cloud migration ETL pipelines and optimized data marts on a cloud-native lakehouse.
- Architected geospatial retail data models and translated analysis into stakeholder reports.
- Delivered ML models for B2B clients, a geospatial feature store, and automated competitive-intelligence scraping.
[cloud migration][etl][sql][geospatial][ml]
Undergraduate Research Fellow
Institute of Biochemistry, Biological Research Centre · Jan 2019 — Jun 2020 · Szeged, Hungary
- Developed computer vision plugins for automated biological image analysis.
- Applied deep learning techniques to high-throughput microscopy data.
- Designed and implemented robust data management systems for large-scale measurements.
[computer vision][deep learning][biomedical imaging]
§4.2Education
MS Business Analytics
Central European University · 2025 — 2026
BS Molecular Bionics
University of Szeged · 2016 — 2020
Exchange — Biomedical Engineering
Universidad CES, Colombia · 2019
§4.3Languages
English — Fluent (C1) · Spanish — Fluent (C1) · German — Basic · Hungarian — Native
§5Certifications & awards
Credentials, badges, and scholarships — each links to the issuing authority where available.
§5.1Certifications
§5.2Awards & scholarships
- Highlighted Study Scholarship of the Hungarian State
- Municipal Scholarship of Szeged
- Campus Mundi Scholarship
- IT Star Award
- Data Rockstar
§6Code
Repositories and side projects — what I've built, shipped, and open-sourced.
§6.1Side projects
Comics Factory
ML-driven SaaS that generates stylized comics with consistent characters from a handful of reference images. Built at Hungary's first hacker space using Supabase, Firebase, and modern image-gen pipelines.
[llms][image gen][saas][supabase]
§6.2Theses
§6.3Open source & experiments
loading repos…