Sebastian Raschka

About Sebastian Raschka

Sebastian Raschka is a Staff Research Engineer at Lightning AI and one of the most influential educators in the machine learning community. He is the author of several widely-used textbooks, including Build a Large Language Model from Scratch and Build a Reasoning Model from Scratch — books that embody the philosophy that the best way to understand AI is to implement it yourself.

Previously an Assistant Professor of Statistics at the University of Wisconsin–Madison (2018–2023), Raschka left academia to focus full-time on LLM research and development at Lightning AI. His work bridges the gap between cutting-edge research and practical implementation, making complex topics accessible through code-driven explanations.

His newsletter Ahead of AI and educational YouTube courses have made him a go-to resource for practitioners who want to understand how LLMs actually work at the implementation level.

Career Highlights

• Staff Research Engineer at Lightning AI (2022–present)
• Former Assistant Professor of Statistics, University of Wisconsin–Madison (2018–2023)
• Author of Build a Large Language Model from Scratch (bestseller)
• Author of Build a Reasoning Model from Scratch
• Author of Machine Learning with PyTorch and Scikit-Learn
• Creator of the Ahead of AI newsletter and educational courses
• 10+ years of experience in AI research and engineering

Notable Positions

On AI Competition in 2026

Raschka argues against the winner-take-all narrative in AI, emphasizing that ideas are no longer proprietary because researchers move between companies and labs. The real competitive advantage is compute budget and hardware access, not secret techniques.

On the Interface Challenge for Agents

Raises the fundamental specification problem for autonomous agents: even if an LLM can execute tasks, how do users communicate complex goals in unstructured environments? Coding works because the environment is well-defined, but general-purpose agents face a much harder interface design problem.

On AI Tool Usage

Describes a pragmatic multi-model approach: ChatGPT's fast mode for quick lookups, Pro mode for thorough document checking, and Gemini for long-context tasks like finding needles in large documents. Uses the Codex VS Code plugin for coding assistance while maintaining control over the process.

Key Quotes

• "I don't think nowadays in 2026 that there will be any company who has access to a technology that no other company has access to. Researchers are frequently changing jobs, changing labs, they rotate." (on AI competition)
• "The problem is for arbitrary tasks, you still have to specify what you want your LLM to do. What is the environment? How do you specify?" (on agents)
• "I use it until it breaks and then I explore other options." (on model switching)

Related Reading

• Scaling Laws - Central topic in his analysis of AI progress
• AI Agents - Discusses the specification challenge
• Pre-training - Explains the ongoing importance of pre-training compute