Tech Teams Today · 2026-05-14 · 40 min
Most conversations about AI right now sound like either the world's ending or it's just beginning. Seng Lin Shee has a different read: we've been here before. In this episode, Seng Lin makes the case that multi-agent systems are essentially multi-threading with English, that good engineering principles don't expire just because the tooling changed, and that the leaders who've already seen a few cycles have most of the mental models they need for this one. We get into how he runs a team that spans interns to 30-year veterans, why he bakes learning directly into performance reviews, and how his team went from building one internal RAG tool to running a platform that helps the rest of a regulated finance company ship GenAI safely. If you've been feeling like the floor is shifting under engineering leadership, this one's a useful counterweight. Seng Lin is one of the more optimistic voices we've had on, and his optimism is grounded in something most people aren't talking about: pattern recognition.
Sr Dir @ OpenGov | How To Cut AI Signal From Noise56 / 100
Dir of Eng @ Mosaic | Over 90% of Their Code Is AI-Generated
Sr Dir @ Twilio | AI, Judgment, and Engineering at Massive Scale
CKIO Fisher Phillips | How a Law Firm Wins Business with Custom AI
CTO Cube | Coding Got Easier. Engineering Didn't.