I’m Max Nardit. I build systems where humans and agents coordinate.
Systems Engineer focused on orchestration, memory, and execution models for AI agents, currently in Bangkok. I work on AI as an operational layer woven into workflows and tools, not as a chatbot.
What I’m interested in
When software stops being fully deterministic and becomes agentic, these questions get interesting:
- how humans coordinate with autonomous systems
- how memory and context evolve beyond single prompts
- how reliability changes when software can reason
- and why orchestration becomes more important as agents get more capable
I focus on the operational reality of these questions: operational complexity, long-lived workflows, observability, local-first environments. Not the speculative version. The running-in-production one.
Day work
I’ve worked at Bobdo, a digital marketing agency in Austria, since 2015. My current focus there is agentic systems for marketing operations: long-running agents around ad campaigns and SEO that handle the work continuously across multiple clients — campaign monitoring on Google and Meta ads, SERP and SEO audits, analytics across GA4 / GSC / ad platforms, report and presentation generation. It’s the real-world version of the agent patterns I think about personally, and it uses every layer of past experience.
Background
Before agents, my background was data engineering, technical SEO, analytics, automation, and infrastructure. That shaped how I think about large interconnected systems, entropy, and reliability at scale. The shift toward agentic systems didn’t feel like a discontinuity. It felt like the next layer of the same problem.
On this site
Writing lives at /articles, short-form notes from real engineering problems. /now is a rotating snapshot of my current focus. /uses is my operational stack. Personal projects (Beetroot, Agent Recall, others) link from the homepage.