Augustin Chan

I spent 10 years at Informatica building and architecting Master Data Management systems for enterprises across Asia and Europe. The kind of work where you learn that boring, reliable infrastructure matters more than clever solutions.

Now I work on AI systems that reason and remember. Mostly the parts that aren't demos: memory architectures that let systems maintain context, domain expertise encoded as structured skills that agents carry and reuse, prompt optimization that's measurable. Postgres for everything. For multi-agent work, I borrow from distributed systems — isolated subagents with skills running in parallel, no shared state, results merged between waves — rather than the “agents coordinating with each other” approach that sounds impressive but rarely works.

I studied Cognitive Science at UC San Diego, where we had more modest goals than trying to build a Digital God. That was a while ago, but the questions are the same.