<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Sushank Ghimire: writing on AI engineering</title><description>Field notes on LLM agents, retrieval systems and production AI backends by Sushank Ghimire.</description><link>https://sushankghimire.com.np/</link><language>en-us</language><item><title>Making tool-calling agents reliable on AWS Bedrock</title><link>https://sushankghimire.com.np/blog/reliable-tool-calling-agents-bedrock-claude/</link><guid isPermaLink="true">https://sushankghimire.com.np/blog/reliable-tool-calling-agents-bedrock-claude/</guid><description>Rate limits, timeout cascades and tool schema drift took down our Claude agents more than any bad prompt did. How we fixed each one and reached 99.9% uptime.</description><pubDate>Wed, 02 Sep 2026 00:00:00 GMT</pubDate><category>agents</category><category>bedrock</category><category>claude</category><category>fastapi</category><author>sushankgghimire@gmail.com (Sushank Ghimire)</author></item><item><title>RAG on messy HTML: preprocessing, structure-aware chunking and hybrid retrieval</title><link>https://sushankghimire.com.np/blog/rag-on-messy-html-chunking-hybrid-retrieval/</link><guid isPermaLink="true">https://sushankghimire.com.np/blog/rag-on-messy-html-chunking-hybrid-retrieval/</guid><description>Fixed-size chunking fails on real documents. What moved retrieval quality for us: boilerplate stripping, heading-aware chunks, hybrid search and a re-ranker.</description><pubDate>Wed, 26 Aug 2026 00:00:00 GMT</pubDate><category>rag</category><category>retrieval</category><category>llm</category><category>python</category><author>sushankgghimire@gmail.com (Sushank Ghimire)</author></item></channel></rss>