I'm Alex, a.k.a Kishin, this is now where I write a little about my projects, labs, and the hassling with AI Agents and other features: evaluation harnesses, retrieval pipelines, guardrails, and the monitoring. No comment section, no newsletter, just posts, written when they're ready.
The reality is, most AI projects don't fail because the model is bad. They fail because nobody built the evaluation, the fallbacks, or the human-in-the-loop escalation path around it. Most of what ends up here comes out of actually building that stuff, not reading about it.
Alongside the writing, the case studies section covers real engagements, and the projects section is a working sample of what's been built. I also take on a small number of consulting engagements on the side, scoped narrowly and measured against a baseline rather than against how good a demo looks.
Feel free to reach out!