Mijo Kristo

10 years inside enterprise software — backends that grew faster than anyone planned, AI that needs to work in production not just in demo, integrations nobody wants to touch. My open-source tools have 120k+ downloads.

Companies come to me when they've tried everything else. Internal teams hit a ceiling. The usual contractors didn't understand the system. What seemed like a straightforward problem keeps getting worse.

The work starts before any code changes. Complex systems accumulate business logic over years — decisions that made sense at the time, constraints nobody documented, behaviour that's never been fully mapped. I work through that systematically: understand the actual requirements, trace what the system does under real conditions, find where reality and the code diverge. That analysis is usually where the problem lives.

I adapt quickly to unfamiliar environments — different stacks, different domains, different levels of technical debt. When AI fits the problem, I integrate it without hesitation: RAG pipelines, LLM orchestration, semantic search across real production data. When it doesn't fit, I say so. Most AI integrations that fail in production aren't AI problems — they're systems problems.

I don't do audits and slide decks. I work inside the system until it's fixed. Two or three engagements a year — if you're at that point, send me an email.