<!DOCTYPE html><html lang="en"> <head> <meta charset="utf-8"> <meta name="viewport" content="width=device-width initial-scale=1" /> <meta http-equiv="X-UA-Compatible" content="IE=edge"> <meta name="author" content="Sebastian Raschka"> <meta property="og:title" content=" Sebastian Raschka "> <meta property="og:description" content=" I'm an LLM Research Engineer with over a decade of experience in artificial intelligence. My work bridges academia and industry, with roles including senior staff at an AI company and a statistics professor. My expertise lies in LLM research and the development of high-performance AI systems, with a deep focus on practical, code-driven implementations. "> <meta property="og:url" content="https://sebastianraschka.com/" /> <meta property="og:site_name" content="Sebastian Raschka, PhD" /> <meta property="og:locale" content="en_US" /> <meta name="twitter:site" content="@rasbt" /> <meta name="twitter:description" content="I'm an LLM Research Engineer with over a decade of experience in artificial intelligence. My work bridges academia and industry, with roles including senior staff at an AI company and a statistics professor. My expertise lies in LLM research and the development of high-performance AI systems, with a deep focus on practical, code-driven implementations." /> <meta property="og:type" content="article" /> <meta property="article:published_time" content="" /> <meta name="twitter:creator" content="@rasbt" /> <meta name="twitter:title" content="Sebastian Raschka" /> <meta name="twitter:card" content="summary_large_image"> <meta name="twitter:image" content="https://sebastianraschka.com/images/logos/hero.jpg"> <title>Sebastian Raschka</title> <meta name="description" content="I'm an LLM Research Engineer with over a decade of experience in artificial intelligence. My work bridges academia and industry, with roles including senior staff at an AI company and a statistics professor. My expertise lies in LLM research and the development of high-performance AI systems, with a deep focus on practical, code-driven implementations."> <link rel="stylesheet" href=" /css/combined_direct_no_sass.css"> <link rel="stylesheet" href=" /css/fork-awesome.min.css"> <meta property='og:title' content="Sebastian Raschka"><meta property="og:type" content="article"><meta property="og:url" content="https://sebastianraschka.com/"> <meta property="og:image" content=""> <meta property="og:description" content=""><meta property="og:site_name" content="Sebastian Raschka, PhD"><meta property="og:locale" content="en_US"> <meta property="fb:admins" content=""><meta property="fb:app_id" content=""> <link rel="canonical" href="https://sebastianraschka.com/"> <link rel="apple-touch-icon" sizes="180x180" href="/images/favicons/apple-touch-icon.png"> <link rel="icon" type="image/png" sizes="32x32" href="/images/favicons/favicon-32x32.png"> <link rel="icon" type="image/png" sizes="16x16" href="/images/favicons/favicon-16x16.png"> <link rel="manifest" href="/site.webmanifest"> <link rel="mask-icon" href="/images/favicons/safari-pinned-tab.svg" color="#5bbad5"> <meta name="msapplication-TileColor" content="#ffc40d"> <meta name="theme-color" content="#ffffff"></head> <body> <img src="/images/logos/ahead-of-ai-icon.png" alt="Ahead of AI logo" style="display: none;"> <header class="site-header"> <div class="site-title" style="text-decoration: none; margin-top: 2em;"> <a href="/"><span style="color:black">Sebastian</span> <span style="color:#c5050c">Raschka</span> </a> <a href="https://x.com/rasbt"><img src="/images/logos/twitter-bw.jpg" height="20" style="padding-left:20px;" alt="Twitter/X icon"></a> <!--<a href="https://threads.net/@sebastianraschka"><img src="/images/logos/threads-logo-alt-small.png" height="20" style="padding-left:5px;" alt="Threads icon"></a>--> <a href="https://www.linkedin.com/in/sebastianraschka/"><img src="/images/logos/linkedin-bw.jpg" height="20" style="padding-left:5px;" alt="LinkedIn Icon"></a> <a href="https://github.com/rasbt"><img src="/images/logos/github-bw.jpg" height="20" style="padding-left:5px;" alt="GitHub icon"></a> </div> <!-- <div style="width:100%;height:50;float:left;margin-bottom:10px;"> <a href="https://twitter.com/rasbt"><img src="/images/logos/twitter-bw.jpg" height="20"></a>--> <!-- <a href="https://www.buymeacoffee.com/rasbt"><img src="/images/logos/coffee-bw.jpg" height="20"></a>--> <!-- <a href="https://mastodon.social/@SebRaschka"><img src="/images/logos/mastodon-bw.jpg" height="20"></a>--> <!-- </div>--> <div class="wrapper"> <nav class="site-nav"> <a href="#" class="menu-icon"> <svg viewBox="0 0 18 15"> <path fill="#424242" d="M18,1.484c0,0.82-0.665,1.484-1.484,1.484H1.484C0.665,2.969,0,2.304,0,1.484l0,0C0,0.665,0.665,0,1.484,0 h15.031C17.335,0,18,0.665,18,1.484L18,1.484z"/> <path fill="#424242" d="M18,7.516C18,8.335,17.335,9,16.516,9H1.484C0.665,9,0,8.335,0,7.516l0,0c0-0.82,0.665-1.484,1.484-1.484 h15.031C17.335,6.031,18,6.696,18,7.516L18,7.516z"/> <path fill="#424242" d="M18,13.516C18,14.335,17.335,15,16.516,15H1.484C0.665,15,0,14.335,0,13.516l0,0 c0-0.82,0.665-1.484,1.484-1.484h15.031C17.335,12.031,18,12.696,18,13.516L18,13.516z"/> </svg> </a> <div class="trigger"> <!--<script type="text/javascript"> var total_images = 2; var random_number = Math.floor((Math.random()*total_images)); var random_img = new Array(); random_img[0] = '<a href="https://twitter.com/rasbt"><img src="/images/logos/twitter-1.jpg" height="20"></a>'; random_img[1] = '<a href="https://linkedin.com/in/sebastianraschka"><img src="/images/logos/linkedin-1.jpg" height="20"></a>'; document.write(random_img[random_number]); </script>--> <span style="padding-left:0px;margin-left:0px;"></span> <a class="page-link" href="https://magazine.sebastianraschka.com"><span style="color:#c5050c;"><img src="/images/logos/ahead-of-ai-icon.png" alt="Ahead of AI Logo" height="20"> AI Magazine</span></a> <!--<a class="page-link" href="/blog/index.html">Blog</a>--> <a class="page-link" href="/books">Books</a> <!--<a class="page-link" href="/newsletter">AI Newsletter</a>--> <!--<a class="page-link" href="/teaching">Courses</a>--> <a class="page-link" href="https://github.com/rasbt/LLMs-from-scratch">LLMs From Scratch</a> <!--<a class="page-link" href="/publications">Research</a>--> <a class="page-link" href="/elsewhere">Talks</a> <a class="page-link" href="/contact">Contact</a> <a class="page-link" href="/resources">More</a> </div> </nav> </div> </header> <div class="page-content"> <div class="wrapper"> <div style="display: flex; align-items: center; justify-content: center; flex-wrap: wrap; max-width: 800px; margin: 40px auto; gap: 30px;"> <div style="flex: 1; min-width: 300px; text-align: left;"> <h1 style="font-size: 1.8em; font-weight: 600; color: #222; margin-bottom: 20px;"> Hello, I'm Sebastian Raschka, PhD </h1> <p style="font-size: 1.1em; line-height: 1.8; color: #555; margin-bottom: 20px;"> I'm an LLM Research Engineer with over a decade of experience in artificial intelligence. My work bridges academia and industry, with roles including senior staff at an AI company and a statistics professor. </p> <p style="font-size: 1.1em; line-height: 1.8; color: #555; margin-bottom: 20px;"> My expertise lies in LLM research and the development of high-performance AI systems, with a deep focus on practical, code-driven implementations. (For my most up-to-date CV details, please visit my <a href="https://www.linkedin.com/in/sebastianraschka">LinkedIn profile</a>.) </p> </div> <div style="flex: 0 0 200px; text-align: center;"> <picture> <source type="image/webp" srcset="/images/logos/photo-2021-08-25_compressed.webp"> <source type="image/jpeg" srcset="/images/logos/photo-2021-08-25_compressed.jpg"> <img src="/images/logos/photo-2021-08-25_compressed.jpg" alt="Sebastian Raschka Profile Picture" style="width: 200px; height: 200px; object-fit: cover; object-position: 100% 10%; border-radius: 50%; box-shadow: 0 4px 10px rgba(0,0,0,0.1);"> </picture> </div></div> <!-- Subscribe Section --><div class="rss" style="display: flex; align-items: center; justify-content: center; width: 96%; margin: 20px auto; max-width: 800px;"> <a href="/rss_feed.xml" style="display: inline-flex; align-items: center; margin-right: 20px;"> <i class="fa fa-rss fa-1x" style="padding-right: 5px;"></i> <span>RSS Feed (Blog and Notes)</span> </a> <a class="page-link" href="https://magazine.sebastianraschka.com/subscribe" style="display: inline-flex; align-items: center;"> <img src="/images/logos/ahead-of-ai-icon.png" alt="Ahead of AI Logo" height="20" style="padding-right: 5px;"> <span>Subscribe via Email (AI Magazine)</span> </a></div> <div style="height: 50px;"></div> <!-- Blog Entries Section --><div style="max-width: 800px; margin: 30px auto;"> <blockquote> <!-- Content about Sebastian --> </blockquote> <div class="post"> <header class="post-header"> <h1 class="post-title">Recent Notes and Blog Entries</h1> </header> <div style="margin-bottom: 45px; text-align: center;"> See <a href="/blog/">Blog and Notes Archive</a> for all entries. </div> <article class="post-content posts-by-year"> <div class="post-entry"> <div class="post-hero-image"> <img src="/images/blog/2025/coding-the-kv-cache-in-llms/hero.jpg" alt="Understanding and Coding the KV Cache in LLMs from Scratch"> </div> <div class="post-details"> <h2 class="post-title"> <a href="/blog/2025/coding-the-kv-cache-in-llms.html">Understanding and Coding the KV Cache in LLMs from Scratch</a> </h2> <p class="post-date">Jun 17, 2025</p> <p class="post-description">KV caches are one of the most critical techniques for efficient inference in LLMs in production. KV caches are an important component for compute-efficient LLM inference in production. This article explains how they work conceptually and in code with a from-scratch, human-readable implementation.</p> </div> </div> <div class="post-entry"> <div class="post-hero-image"> <img src="/images/blog/2025/coding-llms-from-the-ground-up-a-complete-course/hero.jpg" alt="Coding LLMs from the Ground Up: A Complete Course"> </div> <div class="post-details"> <h2 class="post-title"> <a href="/blog/2025/coding-llms-from-the-ground-up-a-complete-course.html">Coding LLMs from the Ground Up: A Complete Course</a> </h2> <p class="post-date">May 10, 2025</p> <p class="post-description">Why build an LLM from scratch? It's probably the best and most efficient way to learn how LLMs really work. Plus, many readers have told me they had a lot of fun doing it.</p> </div> </div> <div class="post-entry"> <div class="post-hero-image"> <img src="/images/blog/2025/the-state-of-reinforcement-learning-for-llm-reasoning/hero.jpg" alt="The State of Reinforcement Learning for LLM Reasoning"> </div> <div class="post-details"> <h2 class="post-title"> <a href="/blog/2025/the-state-of-reinforcement-learning-for-llm-reasoning.html">The State of Reinforcement Learning for LLM Reasoning</a> </h2> <p class="post-date">Apr 19, 2025</p> <p class="post-description">A lot has happened this month, especially with the releases of new flagship models like GPT-4.5 and Llama 4. But you might have noticed that reactions to these releases were relatively muted. Why? One reason could be that GPT-4.5 and Llama 4 remain conventional models, which means they were trained without explicit reinforcement learning for reasoning. However, OpenAI's recent release of the o3 reasoning model demonstrates there is still considerable room for improvement when investing compute strategically, specifically via reinforcement learning methods tailored for reasoning tasks. While reasoning alone isn't a silver bullet, it reliably improves model accuracy and problem-solving capabilities on challenging tasks (so far). And I expect reasoning-focused post-training to become standard practice in future LLM pipelines. So, in this article, let's explore the latest developments in reasoning via reinforcement learning.</p> </div> </div> <div class="post-entry"> <div class="post-hero-image"> <img src="/images/blog/2025/first-look-at-reasoning-from-scratch/hero.jpg" alt="First Look at Reasoning From Scratch: Chapter 1"> </div> <div class="post-details"> <h2 class="post-title"> <a href="/blog/2025/first-look-at-reasoning-from-scratch.html">First Look at Reasoning From Scratch: Chapter 1</a> </h2> <p class="post-date">Mar 29, 2025</p> <p class="post-description">As you know, I've been writing a lot lately about the latest research on reasoning in LLMs. Before my next research-focused blog post, I wanted to offer something special to my paid subscribers as a thank-you for your ongoing support. So, I've started writing a new book on how reasoning works in LLMs, and here I'm sharing the first Chapter 1 with you. This ~15-page chapter is an introduction reasoning in the context of LLMs and provides an overview of methods like inference-time scaling and reinforcement learning. Thanks for your support! I hope you enjoy the chapter, and stay tuned for my next blog post on reasoning research!</p> </div> </div> <div class="post-entry"> <div class="post-hero-image"> <img src="/images/blog/2025/state-of-llm-reasoning-and-inference-scaling/hero.jpg" alt="Inference-Time Compute Scaling Methods to Improve Reasoning Models"> </div> <div class="post-details"> <h2 class="post-title"> <a href="/blog/2025/state-of-llm-reasoning-and-inference-scaling.html">Inference-Time Compute Scaling Methods to Improve Reasoning Models</a> </h2> <p class="post-date">Mar 8, 2025</p> <p class="post-description">This article explores recent research advancements in reasoning-optimized LLMs, with a particular focus on inference-time compute scaling that have emerged since the release of DeepSeek R1.</p> </div> </div> </article> </div></div> <!-- Styles --><style> .post-entry { display: flex; align-items: flex-start; margin-bottom: 20px; border-bottom: 1px solid #ddd; padding-bottom: 20px; } .post-hero-image { flex: 0 0 120px; margin-right: 20px; } .post-hero-image img { width: 100%; height: auto; } .post-details { flex: 1; } .post-title { margin: 0; font-size: 1.5em; color: #333; } .post-title a { color: #333; text-decoration: none; } .post-title a:hover { text-decoration: underline; color: #0056b3; } .post-date { color: #666; font-size: 0.9em; margin: -4px 0; } .post-description { margin: 0; } h1.post-title { text-align: center; } h2.post-title, h2.post-title a { all: unset; display: block; font-size: 1.0em; color: #333; text-decoration: none; } h2.post-title a:hover { color: #0056b3; text-decoration: underline; } a.anchorjs-link { display: none !important; }</style> </div> </div> <footer class="site-footer"> <div class="wrapper"> <div class="footer-col-wrapper"> <div class="footer-col social-col"> <a href="https://magazine.sebastianraschka.com"><span><i class="fa fa-rss fa-2x"></i></span> </a> <a href="/contact"><span><i class="fa fa-envelope fa-2x"></i></span> </a> <a href="https://twitter.com/rasbt"> <span><i class="fa fa-twitter fa-2x"></i></span> </a> <a href="https://youtube.com/c/SebastianRaschka"><span><i class="fa fa-youtube fa-2x"></i></span> </a> <a href="https://github.com/rasbt"><span><i class="fa fa-github fa-2x"></i> </span></a> <a href="https://scholar.google.com/citations?user=X4RCC0IAAAAJ&hl=enrasbt"><span><i class="fa fa-google fa-2x"></i> </span></a> <a href="https://linkedin.com/in/sebastianraschka"><span><i class="fa fa-linkedin fa-2x"></i> </span></a> </div> <div class="footer-col copyright-col"> <p>© 2013-2025 Sebastian Raschka</p> </div> </div></div> <!-- Google tag (gtag.js) --><script async src="https://www.googletagmanager.com/gtag/js?id=G-BYQXBRPK81"></script><script> window.dataLayer = window.dataLayer || []; function gtag(){dataLayer.push(arguments);} gtag('js', new Date()); gtag('config', 'G-BYQXBRPK81');</script></footer></div> <script src="/js/anchor.min.js" type="text/javascript"></script> <script> var selector = 'h2, h3, h4, h5, h6'; /* anchors.options = { icon: '#', visible: 'always', placement: 'left', class: 'bb-anchor' } */ anchors.add(selector); </script> </body></html> 
