About Nojman
Nojman is a one-person ML, MLOps, and GPU-infrastructure studio. We do the work ourselves and let the results carry the argument — proof over claims. Every post here is something we ran, broke, fixed, and measured.
I work hands-on across the whole stack: bare-metal Kubernetes running KubeRay and Ray, 2x RTX 5090 for training and distributed backtesting, PyTorch and LLMs for the models, and an Elixir execution layer where low latency matters. My background is academic — an unfinished PhD that left me comfortable in research-grade math and uncomfortable with hand-waving.
I work bilingually in Slovak and Hungarian and operate into Western-EU teams, which means I read requirements and constraints from both sides of the table — the cost-sensitive Central-European engineering reality and the delivery expectations of Western-EU clients. That is a working advantage, not a line on a map.
Certifications
- AWS Solutions Architect – Professional
- Google Professional Cloud Architect
- CKA — Certified Kubernetes Administrator
- CKS — Certified Kubernetes Security Specialist
- Certified Argo Project Associate
Most of what you read here draws on Raptor, our trading research platform — the body of real, running work the writing is pulled from. It is evidence of what the stack can do, not a product for sale.