Resources & tools
This library brings together tutorials, courses, research, and tools on four core themes in my work: explainable NLP, large language models and their pitfalls, FAIR/anonymized data practices, and causal inference in business & industry.
NLP & XAI
Transformer fundamentals, interpretability methods, and practical tooling for explainable NLP.
- Introduction to BERT & Explainable NLP — Hands-on labs to get started with BERT and common NLP explainability methods.
- Stanford CS224n: NLP with Deep Learning — A graduate-level foundation in modern NLP, from word vectors to transformers.
- DIANNA (Deep Insight and Neural Network Analysis) — Open-source tooling from the Netherlands eScience Center for interpretable ML.
- Personal GitHub (httn22) — My curated NLP/XAI repositories and teaching material on Python for Social Sciences and Causal Inference with NLP.
LLMs, AI & Pitfalls
From coding walkthroughs to societal risks and how to work with LLMs responsibly.
- Transformers (How LLMs Work) Explained Visually — A visual walkthrough of the transformer architecture showing how attention, tokens, and model layers function together; great for building intuition about what happens inside LLMs.
- How I use LLMs (Andrej Karpathy) — Pragmatic workflows and prompts for productive day-to-day use.
- Let’s build GPT from scratch (Andrej Karpathy) — A step-by-step coding deep dive into a small GPT.
- AI Safety should prioritize the Future of Work (Hazra, Majumder & Chakrabarty, 2025) — Why labor-market impacts deserve center stage in AI safety.
FAIR Data Practice & Anonymization Tools
Make datasets reusable and privacy-conscious, with concrete checklists and code.
- FAIR Aware (DANS) — Quick self-assessment of dataset FAIRness.
- Utrecht University Guide to FAIR — Practical steps to make data Findable, Accessible, Interoperable, Reusable.
- Best Practice Anonymization (UK Data Service) — How to anonymize interviews and qualitative text.
- NER-Anon (BERT-based) — Python package for NER-driven text anonymization.
- Textwash — Lightweight anonymizer for Dutch and English text.
Causal Inference
Understand, apply, and integrate causal inference methods with modern machine learning for robust scientific and data-driven reasoning.
- Causal Inference for The Brave and True (by Matheus Facure Alves) — A Python-based open handbook covering causal diagrams, quasi-experiments, and modern causal ML with accessible explanations and code.
- The Effect Book (by Nick Huntington-Klein) — An intuitive online textbook introducing causal effects, why they matter, and how causal thinking differs from descriptive or predictive analysis.
- Applied Causal Inference Powered by ML & AI (Chernozhukov et al., 2024) — A textbook-style guide to combining causal inference with modern ML tools such as double ML, SCMs, and DAGs for high-dimensional applications.
Causal inference provides a rigorous way to go beyond correlations and uncover the true effects of policies, interventions, and business strategies. From measuring advertising ROI to optimizing supply chains and evaluating transport policies, causal methods help organizations make evidence-based decisions. For a deeper dive, see my dedicated page on Causal inference in business and industry.