<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Juergen Neulinger</title><link>https://juergen-neulinger.dev/</link><description>Recent content on Juergen Neulinger</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Mon, 01 Jan 0001 00:00:00 +0000</lastBuildDate><atom:link href="https://juergen-neulinger.dev/index.xml" rel="self" type="application/rss+xml"/><item><title>About</title><link>https://juergen-neulinger.dev/about/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://juergen-neulinger.dev/about/</guid><description>&lt;h3 id="who-am-i">Who am i&lt;/h3>
&lt;p>I&amp;rsquo;m Jürgen Neulinger. I build the infrastructure other people&amp;rsquo;s AI systems run on — agent frameworks, RAG, and the plumbing in between. Right now I lead a generative-AI platform team at Erste Digital, where &amp;ldquo;it works in the demo&amp;rdquo; and &amp;ldquo;it survives an audit&amp;rdquo; are two very different bars. The second one is the job.&lt;/p>
&lt;p>My pattern is to build a foundation to the point a team can own it, then move on to the next one. A few of them: the central capability for provisioning and governing the LLMs used at the bank, now run by the Databricks platform team; a config-driven RAG framework where onboarding a new use case is an index and a little auth config, not new pipeline code; an FSM-based agent framework; and an MCP gateway with real enterprise auth — Entra ID, per-tool OPA/Rego authorization, OAuth 2.1 On-Behalf-Of.&lt;/p></description></item></channel></rss>