Building ML systems 2025-10-03T00:00:00.000Z https://example.com/ Ramesh Arvind 2025-09-28T00:00:00.000Z https://example.com/posts/cloudflare-appreciation/ Cloudflare continues to impress. They operate a meaningful slice of the modern web, but what stands out to me is how consistently polished many of their features are. I don’t use every Cloudflare product, just an appreciation for good engineering and thoughtful UX. A recent example: I bought a new domain and tried Cloudflare’s Email Routing. It was a seamless setup took just a few clicks, and it worked immediately. I’m looking forward to exploring more of their stack over time. When infrastructure fades into the background, that’s when you know the product is doing its job.

A short note appreciating Cloudflare—especially how seamless Email Routing is.

2025-10-03T00:00:00.000Z https://example.com/posts/prompt-engineering-with-ai/ Models are very good at writing and improving prompts — they’re a helpful first pass, not the final word. Treat the model like a fast collaborator: give it enough context to understand the task, let it propose small changes, then you decide what sticks. Checklist and workflow: Start with 1–2 AI revisions as a first pass, then evaluate on real examples; you decide what sticks. Provide real context and concrete feedback; avoid generic requests like “Improve the prompt” or “Reduce hallucinations.” Include 1–3 representative I/O examples with brief notes to guide meaningful edits. Ask for targeted edits on exact passages; keep changes small and focused. Use eval tooling when useful (e.g., LLM‑as‑a‑judge), but don’t overfit—prefer minimal, generalizable changes. Distill many feedback items into concise guidance before edits if needed; evaluate changes on real data, since automated evals can be wrong. If AI revisions don’t help, perform manual edits to set direction, then bring the model back for fine‑tuning. Avoid more than two consecutive AI‑only edits to prevent drift and overfitting. Expect prompts to grow; prioritize clarity and useful context first, then condense when length truly matters. When you follow this approach, model‑assisted prompt editing becomes a force‑multiplier!

Practical guidelines for using models to write and improve prompts.