<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>ductai05</title><description>Personal portfolio of Duc Tai Dinh — Backend Engineer, AI Researcher, and AI student in Ho Chi Minh City.</description><link>https://ductai05.github.io/</link><item><title>From AI Research to Production: Optimizing Multimodal Retrieval, OCR, and System Design</title><link>https://ductai05.github.io/blog/from-research-to-production-ai/</link><guid isPermaLink="true">https://ductai05.github.io/blog/from-research-to-production-ai/</guid><description>A grounded technical breakdown of three multimodal AI systems I co-built in 2025 — KPTER (SOICT 2025), ZSE-Cap (Top-4 @ ACM MM 2025), and a fine-tuned Sino-Nom OCR detector — and the engineering lessons that survived past the leaderboard.</description><pubDate>Tue, 05 May 2026 00:00:00 GMT</pubDate></item><item><title>Blog Post Template — All Features Showcase</title><link>https://ductai05.github.io/blog/template/</link><guid isPermaLink="true">https://ductai05.github.io/blog/template/</guid><description>A comprehensive MDX template demonstrating every available feature: callouts, code blocks with diffs, math, tables, images, and more.</description><pubDate>Thu, 01 Jan 2026 00:00:00 GMT</pubDate></item></channel></rss>