Operator level thinking on AI engineering, embedded delivery, and what actually ships in mid market production environments. Two long form posts a month. No fluff, no marketing.
The things our engineers wish their last team had known before the AI pilot stalled. Patterns from our own work, drawn directly from active engagements.
What works when the corpus is messy, the queries are weird, and the customer expects an answer in 800ms. Hybrid retrieval, reranking, eval harnesses, the unglamorous parts.
What a real US senior costs all in. What an embedded LATAM senior costs all in. The hidden lines on both. Why the comparison usually leaves out tenure, which is the line that matters most.
Field notes from the framework. Why the diagnostic phase is paid. Why the 30/60/90 has the shape it has. What goes wrong when teams skip the match phase.
Not Series A speculation, not Fortune 100 procurement theater. Real AI initiatives at the size of company where the budget is real and the constraints are too.
Drop your email. Unsubscribe anytime. We will never share or sell the list.
Federico Ramallo's long form writing, the PreVetted Podcast, and the upcoming book The Invisible Distance all live at his personal site. The personal assets live there. The company assets live here.
Both compound. If you want the operator's voice rather than the company's, that is the place to start.