<?xml version="1.0" encoding="UTF-8"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Wasim Jalali · Applied AI Engineer</title><description>Applied AI engineer in Germany. I build production AI products end to end with AI coding agents: full-stack apps, internal tools and agent systems.</description><link>https://wasimjalali.com/</link><language>en-gb</language><atom:link href="https://wasimjalali.com/rss.xml" rel="self" type="application/rss+xml"/><item><title>Changing the AI model cut made-up claims in job applications</title><link>https://wasimjalali.com/writings/from-24-made-up-claims-to-1-changing-the-model-behind-job-applications/</link><guid isPermaLink="true">https://wasimjalali.com/writings/from-24-made-up-claims-to-1-changing-the-model-behind-job-applications/</guid><description>Asked about skills an applicant didn&apos;t have, the old model made up a claim in 24 of the 33 answers a strict checker read. Claude Haiku 5.5 did it once in 29.</description><pubDate>Fri, 23 Oct 2026 00:00:00 GMT</pubDate><author>hello@wasimjalali.com (Wasim Jalali)</author></item><item><title>How I made a live-chat helper almost 8 times lighter</title><link>https://wasimjalali.com/writings/7-8-times-less-cpu-a-live-chat-question-filter-at-400-comments/</link><guid isPermaLink="true">https://wasimjalali.com/writings/7-8-times-less-cpu-a-live-chat-question-filter-at-400-comments/</guid><description>A Chrome extension that sorts viewers&apos; questions was slowing the teacher&apos;s laptop. Seven rounds of fixes cut its work 7.8 times.</description><pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate><author>hello@wasimjalali.com (Wasim Jalali)</author></item><item><title>Why I threw away the fix that scored perfectly</title><link>https://wasimjalali.com/writings/from-116-to-118-of-120-the-fix-i-threw-away/</link><guid isPermaLink="true">https://wasimjalali.com/writings/from-116-to-118-of-120-the-fix-i-threw-away/</guid><description>A company Q&amp;A assistant now gets 118 of 120 questions right, with a fix that isn&apos;t tied to the test&apos;s own documents.</description><pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate><author>hello@wasimjalali.com (Wasim Jalali)</author></item><item><title>How I stopped an AI from hiding real questions</title><link>https://wasimjalali.com/writings/from-41-hidden-questions-to-zero-making-a-live-qa-filter-safe-to-trust/</link><guid isPermaLink="true">https://wasimjalali.com/writings/from-41-hidden-questions-to-zero-making-a-live-qa-filter-safe-to-trust/</guid><description>A filter for a live Q&amp;A used to hide 41 of 618 real questions. A second, careful check brought that to zero.</description><pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate><author>hello@wasimjalali.com (Wasim Jalali)</author></item><item><title>Replaying a live broadcast to fix a confusing bug</title><link>https://wasimjalali.com/writings/from-six-confusing-cards-to-none-fixing-a-live-qa-filter-by-replay/</link><guid isPermaLink="true">https://wasimjalali.com/writings/from-six-confusing-cards-to-none-fixing-a-live-qa-filter-by-replay/</guid><description>Six cards on screen made no sense to the teacher. Replaying the broadcast found why, and the ninth replay got all 115 comments right.</description><pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate><author>hello@wasimjalali.com (Wasim Jalali)</author></item><item><title>I tested 28 AI models, and the test picked the winner</title><link>https://wasimjalali.com/writings/wrongly-hidden-questions-4-0-to-1-9-percent-the-test-design-chose-the-model/</link><guid isPermaLink="true">https://wasimjalali.com/writings/wrongly-hidden-questions-4-0-to-1-9-percent-the-test-design-chose-the-model/</guid><description>How the way you test changes which model looks best, and how wrongly hidden questions fell from 4.0% to 1.9% on the test.</description><pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate><author>hello@wasimjalali.com (Wasim Jalali)</author></item><item><title>The cheaper AI model did the same job for a ninth of the price</title><link>https://wasimjalali.com/writings/picking-a-model-for-a-grounded-rag-agent/</link><guid isPermaLink="true">https://wasimjalali.com/writings/picking-a-model-for-a-grounded-rag-agent/</guid><description>Five models took the same 120-question test. The cheap one tied the expensive one at 95%.</description><pubDate>Tue, 01 Sep 2026 00:00:00 GMT</pubDate><author>hello@wasimjalali.com (Wasim Jalali)</author></item><item><title>Teaching an assistant to answer only from its sources</title><link>https://wasimjalali.com/writings/from-72-to-95-repairing-a-grounded-rag-agent/</link><guid isPermaLink="true">https://wasimjalali.com/writings/from-72-to-95-repairing-a-grounded-rag-agent/</guid><description>From 72% to 95% correct answers, without changing how the answers were graded.</description><pubDate>Tue, 01 Sep 2026 00:00:00 GMT</pubDate><author>hello@wasimjalali.com (Wasim Jalali)</author></item></channel></rss>