Hey, I’m Feđa. I live in Tuzla, Bosnia and Herzegovina, and I’ve spent the last 10+ years going from writing my first Android app to leading engineering teams and, these days, getting genuinely obsessed with what AI agents can do when you let them loose on real problems. This blog is where I document all of it: the wins, the dead ends, and the “wait, that actually worked?” moments.

What I do

I’m an Engineering Manager at Personify Health, where I lead a cross-functional team of full stack, mobile, and QA engineers working on backend and platform initiatives like service migrations, observability upgrades, and system reliability. I partner closely with product managers, tech leads, and other teams to ship reliable healthcare technology.

Day to day, my job is a mix of people leadership, delivery ownership, and technical alignment. I help engineers grow, keep priorities clear, and make sure the work we do actually moves the product forward, not just the backlog.

Before moving into management I spent years hands-on, and honestly I still try to stay that way. Most recently I was Mobile Tech Lead, where I drove our Clean Architecture migration and company-wide mobile modularisation, set up SRE reporting and platform health monitoring, and leaned hard into AI tooling like Claude Code and Windsurf to speed up development and automate monitoring and audit work with Python. That shift, watching an agent competently handle work that used to eat a full afternoon, is a big part of why I can’t stop building with this stuff outside of work either.

What I build outside work

Outside of work I’m deep into AI-driven development: agents, local LLMs, and self-hosting. I like problems with a clear “does this actually work” test at the end, which is probably why so many of my side projects end up being multi-agent systems that either produce something useful or fall over in interesting ways. I also run most of my own infrastructure, partly for privacy, partly because I just enjoy owning the whole stack from the VPS up. Some of it lives in the Projects section:

  • Site Analytics Tool, a self-hosted analytics control center combining GA4, Search Console, and Bing data.
  • SplitDecision, four AI agents with distinct personalities debating your decisions in real time.
  • Utility Forge, an autonomous 3-agent pipeline that ships a new developer tool every day using GitHub as its runtime.
  • PocketLLM, Gemma 3n running fully offline on Android with MediaPipe.

The posts go deeper: running LLMs locally, MCP servers, multi-agent workflows, and self-hosting on a VPS with Coolify.

The path here

I started out freelancing from 2012 to 2015, then moved into Android and Java backend development at CodeNest in Tuzla. At ABC Tech Group I led Kotlin migrations and worked on an Ethereum mobile wallet and a mobile banking app. In 2019 I joined Virgin Pulse (now Personify Health) as an Android engineer. From there I grew into Mobile Team Lead, running a four-person team and owning releases, CI/CD pipelines, mentoring, and a mobile bootcamp program. After that I became Mobile Tech Lead for the cross-platform iOS and Android team, and today I’m an Engineering Manager on the backend and platform side.

Along the way my toolbox grew from Android and Kotlin into Kotlin Multiplatform, Python, backend systems, and everything AI: agents, prompt engineering, on-device models, and the infrastructure to run it all myself.

I hold a Bachelor of Computer Science from the Faculty of Information Technology, International University Travnik.

Beyond the screen

When I’m not shipping code or arguing with an AI agent about why its tests are failing, I’m usually still thinking about the next thing to build. I genuinely enjoy the process of taking a rough idea, poking at it until it’s a working prototype, and then writing up what I learned so the next person (often future me) doesn’t have to rediscover it the hard way.

Get in touch

Always happy to talk engineering leadership, AI agents, or self-hosting.