Make it
reliable.

I'm Fabijan — a platform/SRE engineer who makes AI systems work in production. Nine years building cloud-native platforms across banking, defense, healthcare, blockchain and e-commerce; now focused on the infrastructure behind reliable agentic AI.

Approach

Take complex distributed systems and make them boring.

The pattern across banking, defense, healthcare and blockchain has been the same: make failure modes explicit, bind trust to evidence, and keep the thing running long after everyone's moved on. Now I'm applying that production muscle to agentic AI — RAG serving, LLM deployment across environments, and MCP-native tooling.

Career

IBM

8+ yrs · two technical achievement awards

Production engineering across nine organizations and nine industries — from e-commerce to a cross-project reliability lead role to the infrastructure behind production AI.

CWI

Distributed-systems research at Centrum Wiskunde & Informatica; co-authored an ACM Multimedia paper on video sharing over software-defined networks.

JogoJogo

First engineering role. Promoted to project lead within four months; ran an 8-person team building mobile apps.

Interests

What the work keeps returning to

  • Binding agent permissions to evidence, not environments
  • Day-2 platform problems — keeping systems efficient after launch
  • MCP-native tooling and the operational substrate for running agents