---
title: "RYNAUT (@rynaut) | Hashnode"
description: "RYNAUT (@rynaut) writes on Hashnode. Read their latest articles and follow their work."
url: "https://hashnode.com/@rynaut"
---

@rynaut

# RYNAUT

@rynaut · Bangalore, India · Joined August 2026

### [Why Your AI Agent Charged a Customer Twice: Idempotency Explained](https://rynaut.hashnode.dev/why-your-ai-agent-charged-the-customer-twice?utm_source=hashnode&utm_medium=feed)

Why do AI agent retries cause duplicate payments? Learn how idempotency keys, atomic transactions and concurrency controls prevent duplicate charges.

4 min read · [rynaut.hashnode.dev](https://rynaut.hashnode.dev/why-your-ai-agent-charged-the-customer-twice?utm_source=hashnode&utm_medium=feed)

[Oct 18](https://hashnode.com/posts/why-your-ai-agent-charged-the-customer-twice/6ad444b832882ddf38cc0cf5)

### [Google’s Own SRE Book Explains Why Your Incident-Response AI Should Never Touch the Pager](https://rynaut.hashnode.dev/google-s-own-sre-book-explains-why-your-incident-response-ai-should-never-touch-the-pager?utm_source=hashnode&utm_medium=feed)

At 3 AM, an alerting system isn’t a dashboard — it’s the one thing standing between a degraded service and an unattended outage. So it’s no surprise that incident response looks like an obviously good

5 min read · [rynaut.hashnode.dev](https://rynaut.hashnode.dev/google-s-own-sre-book-explains-why-your-incident-response-ai-should-never-touch-the-pager?utm_source=hashnode&utm_medium=feed)

[1](https://hashnode.com/posts/google-s-own-sre-book-explains-why-your-incident-response-ai-should-never-touch-the-pager/6a8860c7a04b7c234196d833) [Aug 21](https://hashnode.com/posts/google-s-own-sre-book-explains-why-your-incident-response-ai-should-never-touch-the-pager/6a8860c7a04b7c234196d833)

### [Netflix Solved Adaptive Rate Limiting With TCP Congestion Control. Not With a Model.](https://rynaut.hashnode.dev/netflix-solved-adaptive-rate-limiting-with-tcp-congestion-control-not-with-a-model?utm_source=hashnode&utm_medium=feed)

Rate limiting looks like exactly the kind of decision that should belong to a model. A static limit — some fixed number of concurrent requests — is genuinely bad, and bad in a specific way: it needs c

6 min read · [rynaut.hashnode.dev](https://rynaut.hashnode.dev/netflix-solved-adaptive-rate-limiting-with-tcp-congestion-control-not-with-a-model?utm_source=hashnode&utm_medium=feed)

[Aug 9](https://hashnode.com/posts/netflix-solved-adaptive-rate-limiting-with-tcp-congestion-control-not-with-a-model/6a77d9e87c75163941b3e312)

### [Your AI Coding Agent Isn't Making Software Worse By Failing. It's Making Software Worse By Passing](https://rynaut.hashnode.dev/your-ai-coding-agent-isn-t-making-software-worse-by-failing-it-s-making-software-worse-by-passing?utm_source=hashnode&utm_medium=feed)

"Passing tests" has never meant "correct." Every engineer who's shipped a bug wrapped in green checkmarks knows this. What's changed isn't the existence of the gap — it's who's exploiting it, delibera

4 min read · [rynaut.hashnode.dev](https://rynaut.hashnode.dev/your-ai-coding-agent-isn-t-making-software-worse-by-failing-it-s-making-software-worse-by-passing?utm_source=hashnode&utm_medium=feed)

[Aug 1](https://hashnode.com/posts/your-ai-coding-agent-isn-t-making-software-worse-by-failing-it-s-making-software-worse-by-passing/6a6dd6116e8c9bd54dd899bc)