If you are building an agent, packing the request is part of the product. Short rules, tools on demand, a stable cache prefix, and cost measured per finished task. Same work. Less waste. lnkd.in/p/dfcbHShW

Show Post

Don’t try to build everything first. Ship a useful step, learn, then the next one. Big work is a marathon of small outcomes not one long silence.

Show Post

My take on getting out: work in increments, and make each increment outcome-driven. Not weeks or months of building everything behind a closed door. A small slice a consumer can actually use. That is how you get speed and early feedback on what is worth pitching next.

Show Post

As a senior engineer or any engineer: have you felt the "senior engineer death spiral"? You take on something bigger to prove yourself, disappear for weeks, keep saying "going well," and quietly panic because nothing has shipped.

Show Post

- Route a prompt to the right-sized model - Block injection or input before it hits the LLM - Allow, ask, or deny a tool call - Triage an inbox - Rerank passages - Score an eval - Label a huge table - Run a control loop every few hundred milliseconds - Hand off to a human when confidence is low

Show Post