oxykodit / sofie-van-landeghem## Background
- Back in 2007 we just called it Machine Learning and Big Data.
+ A PhD in Natural Language Processing, and a career in AI ever since.
+ I love coding, debugging, documenting, testing, nitpicking and fine-tuning.
+ I prefer code to be sufficiently robust and modular,
+ and I care a great deal about user experience (UX) and quality assurance (QA).
maintains
FastAPI, Typer & SQLModel — and recent work on Karpathy's nanochat and autoresearch.
archived
spaCy, spacy-transformers, spacy-llm and Thinc at Explosion; HuggingFace's neuralcoref (2019–2023).
writes
The Diff & The Merge — the day-to-day of being an OSS maintainer.
OS repo's I maintain & contributed to
github/svlandeg ↗karpathy/nanochat
The best ChatGPT that $100 can buy.
Python★ 56.9k⑂ 7.9kkarpathy/autoresearch
AI agents running research on single-GPU nanochat training automatically
Python★ 92.8k⑂ 13.2kfastapi/fastapi
FastAPI framework, high performance, easy to learn, fast to code, ready for production
Python★ 101.2k⑂ 9.7kfastapi/typer
Typer, build great CLIs. Easy to code. Based on Python type hints.
Python★ 19.9k⑂ 965fastapi/sqlmodel
SQL databases in Python, designed for simplicity, compatibility, and robustness.
Python★ 18.2k⑂ 881explosion/spaCy
💫 Industrial-strength Natural Language Processing (NLP) in Python
Python★ 33.8k⑂ 4.7k
Talks, Events & Publications
Full archive (36) →Conference talks, keynotes, publications and older posts.
- 1 Apr 2026Nanochat: training your own LLM chat bot from scratch
- 16 Oct 2025How to not get fooled by your data while AI engineering
- 11 Jun 2025What open-weight LLMs today can (and can't) do
- 31 May 2025Data doesn't lie — but it can mislead: How to ensure integrity of your Machine Learning applications
- 12 Mar 2025Gemma 3 open-weights models
- 28 Jun 2024Once a Maintainer: Sofie Van Landeghem
The Diff & The Merge
Substack ↗The day-to-day life of an open-source maintainer in the Python / data / AI domain.
