I started as a software developer in 2006 and later moved through CTO, operations, and managing-director roles. Today I lead Computacenter Romania. Alongside that role, I am a founding partner and CEO of N13 Ventures and pursue a PhD in artificial intelligence at Babeș-Bolyai University.
This profile is the hands-on record: small Romanian language models, public emulator fixes, reproducible machine configurations, and the tests behind them.
- 86Box merged contributions — upstream fixes in memory management, input, threading, printer and audio emulation, Windows pipe I/O, and macOS platform code.
- 86Box VM Recipes — versioned, media-free machine configurations with checks for scripts, manifests, and the public/private file boundary.
- Retro Hardware Lab — later-PC emulation experiments organized around questions, stop conditions, short reports, and decision records.
- Engineering notes — longer accounts of public systems work, including three merged emulator fixes and their tests.
My doctoral work follows controlled narratives from English generation through English–Romanian translation to compact Romanian models trained from scratch. A parallel line examines Romanian diacritic restoration and open-weight models as evaluators.
- Synthetic Data Generation Using Large Language Models: Advances in Text and Code — a survey of LLM-generated text and code datasets, published in IEEE Access.
- Building Large-Scale English–Romanian Literary Translation Resources with Open Models — parallel literary data, fine-tuned open models, and evaluation for English–Romanian translation.
- TF3-RO-50M — Romanian tokenizers and a compact language model trained from scratch, then compressed for local use.
- TF1-EN-3M — three million synthetic moral fables for studying small, open language models.
The research map connects the papers and released artifacts. The full publication list is on Google Scholar and ORCID.
Teaching material and smaller experiments live in notebooks.
Blog · Google Scholar · ORCID · LinkedIn