Builders & open source
Sebastian Raschka
LLM research engineer and author of Build a Large Language Model (From Scratch); formerly at Lightning AI
Sebastian Raschka is an LLM research engineer whose recent work, according to Raschka's site, centers on LLM architectures and small-scale training, with an interest in inference and evaluation. Raschka is the author of Build a Large Language Model (From Scratch) and its sequel, Build a Reasoning Model (From Scratch); the first book's official code repository implements, pretrains and fine-tunes a GPT-like model in PyTorch and is designed to run on conventional laptops. Raschka's site describes earlier roles as a senior engineer at Lightning AI and as a statistics professor at the University of Wisconsin-Madison.123
At a glance
- Current role
- None confirmed in the sources
- Sector
- Independent
- Work listed
- 1
- Sources cited
- 7
- Profile reviewed
Career
Raschka was a professor in the Department of Statistics at the University of Wisconsin-Madison, teaching statistics, machine learning and deep learning and doing research on deep neural networks. Raschka later worked as a senior engineer at Lightning AI and now describes the current work as that of an LLM research engineer, focused on LLM architectures and small-scale training.21
Raschka's research publications include work on rank-consistent ordinal regression with deep neural networks and machine learning classifiers for predicting protein–protein interaction hot spots, and a 2026 preprint with Sten Rüdiger comparing a fine-tuning method called MiCA with LoRA and full fine-tuning.4
Teaching and writing
Raschka's book Python Machine Learning was published by Packt in 2015 and followed by second and third editions with Vahid Mirjalili, and by Machine Learning with PyTorch and Scikit-Learn in 2022 with Yuxi (Hayden) Liu and Mirjalili. In 2024 Raschka published Machine Learning Q and AI with No Starch Press and Build a Large Language Model (From Scratch) with Manning.45
The Manning sequel, Build a Reasoning Model (From Scratch), starts from a pretrained model and implements inference-time scaling, reinforcement learning training and distillation in code. Each of the from-scratch books has an official code repository on GitHub, and Raschka's site also hosts a blog, course materials and an LLM Gallery.5631
Work and contributions
LLMs-from-scratch Open-source code
Official code repository for Raschka's book Build a Large Language Model (From Scratch), containing code for developing, pretraining and fine-tuning a GPT-like LLM in PyTorch without external LLM libraries; the main chapter code is designed to run on conventional laptops. Licensed under Apache-2.0.371
Timeline
- Published Python Machine Learning with Packt Publishing.4
- Published the second edition of Python Machine Learning with Vahid Mirjalili.4
- Published the third edition of Python Machine Learning with Vahid Mirjalili.4
- Published Machine Learning with PyTorch and Scikit-Learn with Packt Publishing.4
- Published Machine Learning Q and AI with No Starch Press.4
- Published Build a Large Language Model (From Scratch) with Manning Publications.4
Documented roles
The sources do not confirm a current role; these are the roles they document.
- Senior engineer, Lightning AI12
- Statistics professor, University of Wisconsin-Madison12
Companies in their history
- Lightning AI 3 people in the wiki
Related people
- William FalconConnected to Lightning AI
- Luca AntigaConnected to Lightning AI
- Alexander AminiLanguage models · Teaching and writing
- Rowan CheungTeaching and writing · Language models
- AI ExplainedTeaching and writing · Language models
- Alessio FanelliTeaching and writing · Language models
- Sasha RushLanguage models · Teaching and writing
- Dan JurafskyLanguage models · Teaching and writing
- Nathan LambertLanguage models · Teaching and writing
- Dwarkesh PatelTeaching and writing · Language models
Suggest a correction or removal
Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.
References
7 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026