<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Blog on Chetan Dalal — Senior Software Engineer</title><link>https://chetandalal.net/posts/</link><description>Recent content in Blog on Chetan Dalal — Senior Software Engineer</description><generator>Hugo -- gohugo.io</generator><language>en-US</language><copyright>© 2019–2026 Chetan Dalal</copyright><atom:link href="https://chetandalal.net/posts/index.xml" rel="self" type="application/rss+xml"/><item><title>Evaluating an AI Feature When There's No Ground Truth</title><link>https://chetandalal.net/posts/evaluating-ai-without-ground-truth/</link><pubDate>Sat, 01 Aug 2026 00:00:00 +0000</pubDate><guid>https://chetandalal.net/posts/evaluating-ai-without-ground-truth/</guid><description>Most AI features ship without a defensible way to tell whether a change made them better. Here&amp;rsquo;s how to build an evaluation approach when ground truth doesn&amp;rsquo;t exist, and how to be honest about what it can&amp;rsquo;t tell you.</description></item><item><title>Deleting Alerts Made Us More Reliable Than Adding Them Ever Did</title><link>https://chetandalal.net/posts/deleting-alerts-improved-reliability/</link><pubDate>Wed, 08 Jul 2026 00:00:00 +0000</pubDate><guid>https://chetandalal.net/posts/deleting-alerts-improved-reliability/</guid><description>The instinct after an incident is to add an alert. The higher-value move is usually deleting three that have never led to action — and here&amp;rsquo;s how to make that argument without losing it.</description></item><item><title>Retries Are the Most Misused Resilience Pattern</title><link>https://chetandalal.net/posts/retries-are-misused/</link><pubDate>Wed, 10 Jun 2026 00:00:00 +0000</pubDate><guid>https://chetandalal.net/posts/retries-are-misused/</guid><description>Retries are the first resilience pattern everyone adds and the one most likely to cause the outage it was meant to prevent. The mechanics of retry storms, and how to bound them.</description></item><item><title>Most Production Incidents Are Data Problems in Disguise</title><link>https://chetandalal.net/posts/data-problems-in-disguise/</link><pubDate>Wed, 13 May 2026 00:00:00 +0000</pubDate><guid>https://chetandalal.net/posts/data-problems-in-disguise/</guid><description>The incident looks like an application bug for the first hour. Why data failures propagate silently, why they cost more than outages, and what treating data quality as an availability concern changes.</description></item><item><title>The Observability Bill Nobody Warns You About: Cardinality</title><link>https://chetandalal.net/posts/observability-cardinality-cost/</link><pubDate>Wed, 15 Apr 2026 00:00:00 +0000</pubDate><guid>https://chetandalal.net/posts/observability-cardinality-cost/</guid><description>Every label you add to a metric multiplies its cost. The failure mode is gradual, invisible, and ends in a budget review — here&amp;rsquo;s how to decide what not to measure.</description></item><item><title>Hello, world</title><link>https://chetandalal.net/posts/hello-world/</link><pubDate>Wed, 25 Mar 2026 00:00:00 +0000</pubDate><guid>https://chetandalal.net/posts/hello-world/</guid><description/></item></channel></rss>