<?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>c++ on AI Logs</title><link>https://drafts.ragv.in/tags/c++/</link><description>Recent content in c++ on AI Logs</description><generator>Hugo</generator><language>en-us</language><copyright>2025 Raghava Dhanya · License</copyright><lastBuildDate>Thu, 30 Jun 2022 16:54:09 +0530</lastBuildDate><atom:link href="https://drafts.ragv.in/tags/c++/index.xml" rel="self" type="application/rss+xml"/><item><title>Python with a Dash of C++: Optimizing Recommendation Serving</title><link>https://drafts.ragv.in/posts/python-with-a-dash-of-cpp-optimizing/</link><pubDate>Thu, 30 Jun 2022 16:54:09 +0530</pubDate><guid>https://drafts.ragv.in/posts/python-with-a-dash-of-cpp-optimizing/</guid><description>&lt;p&gt;Serving recommendation to 200+ millions of users for thousands of candidates with less than 100ms is &lt;strong&gt;hard&lt;/strong&gt; but doing that in Python is &lt;strong&gt;harder&lt;/strong&gt;. Why not add some compiled spice to it to make it faster? Using Cython you can add C++ components to your Python code. Isn&amp;rsquo;t all machine learning and statistics libraries already written in C and Cython to make them super fast? Yes. But there&amp;rsquo;s still some optimizations left on the table. I&amp;rsquo;ll go through how I optimized some of our sampling methods in the recommendation system using C++.&lt;/p&gt;</description></item></channel></rss>