Today I Learned

A daily log of things I actually needed to look up, figure out, or didn't know yesterday. No polished tutorials — raw notes, honest gaps, real problems solved.

---

Format

Each entry is a markdown file named DD-topic-slug.md inside til/YYYY/month/.

til/
├── 2026/
│   ├── avril/
│   │   ├── 16-git-rebase-vs-merge.md
│   │   ├── 17-docker-multistage.md
│   │   └── ...
│   └── mai/
│       └── ...

---

Archive

May 2026

Date	Topic	Tags
May 02	dbt — why it exists and the 4 concepts that matter most	#data #sql
May 01	Vector databases — what they actually do vs a regular DB	#ai #data

April 2026

Date	Topic	Tags
Apr 30	ESG data quality — the 5 dimensions that actually matter for CSRD	#esg #data
Apr 29	GitHub Actions — minimal CI for a Python project	#git #mlops
Apr 28	Python async/await — the mental model that finally made it click	#python #fastapi
Apr 27	RAG chunking strategies — why naive splitting breaks retrieval	#ai #llm #rag
Apr 26	Git hooks with pre-commit — automated code quality without thinking	#git #python
Apr 25	FastAPI background tasks vs Celery — when to use which	#python #fastapi
Apr 24	SQL window functions — the 4 I use most	#sql #data
Apr 23	Scope 3 Category 15 — why financed emissions are the hardest	#esg #ghg
Apr 22	Pandas memory optimization — category dtype and chunked reads	#python #data
Apr 21	Python dataclasses vs Pydantic — which one when	#python
Apr 20	How transformer attention actually scales memory with sequence length	#ml #transformers
Apr 19	BigQuery UNNEST() on repeated fields	#bigquery #sql
Apr 18	CSRD double materiality — what "financial" vs "impact" actually means	#esg #csrd
Apr 17	Docker multi-stage builds cut image from 1.2GB → 180MB	#docker #mlops
Apr 16	Git rebase vs merge — when each actually makes sense	#git