Writing Guide

All notes live in _posts/ and are filtered by category (content type). Projects have permanent hubs in _projects/.

Rule of thumb: category = type of note · domain = subject area · project_id = link to a project · tags = technologies and search keywords.

Content types (categories)

Category Purpose Listing Template
blog Industry news, opinions, commentary /writing/ _templates/blog-post.md
progress Project progress journal /writing/ _templates/progress-note.md
learning Guided exercises, coursework, tutorials /writing/ _templates/learning-note.md
explainer Deep dive on a concept /writing/ _templates/explainer-note.md
challenge Kaggle, ENS, hackathon write-ups /writing/ _templates/challenge-note.md
idea Quick future project ideas /writing/ _templates/idea-note.md
reading Books, papers, courses /writing/ _templates/reading-note.md

Which category to use?

You are writing… Category Key front matter
Actu / opinion piece blog domain
Project milestone update progress project_id, domain, status, stage
Guided exercise (“projet bateau”) learning domain, autonomy: exercise
Weekend exploration learning or progress autonomy: little-experiment
Future project idea (quick note) idea domain
Future project idea (serious) hub in _projects/ status: goal (+ optional idea post)
Concept deep dive explainer domain
Kaggle / ENS write-up challenge domain, competition

Shared front matter

Use on any note where relevant (not as categories):

  • domain — data-science, geospatial, mlops, web-dev, ai-agents, optimization, energy-modeling
  • autonomy — exercise, little-experiment, solo-project
  • stage — notebook, prototype, mvp, app
  • status — goal, wip, done (mainly progress and project hubs)
  • project_id — slug linking notes to /projects/your-id/
  • competition — slug grouping challenge posts (e.g. titanic, ens-challenge-2024)

See the Project Taxonomy Guide for field details.

Tags

Technologies and free-form keywords only: Python, PyTorch, Kaggle, time-series. Do not use tags for content type or domain.

Note structures

Blog

Clear sections, link related projects or reading notes.

Progress

  1. Link to project hub
  2. What you did
  3. Next steps

Learning

  1. Context (course / tutorial source)
  2. What you did
  3. Takeaways

Explainer

  1. Problem
  2. Intuition
  3. How it works
  4. Example
  5. When to use it

Challenge

  1. Competition overview
  2. Approach
  3. Results
  4. Lessons learned

Idea

  1. Motivation
  2. Rough plan
  3. Open questions

Reading

  1. Overview
  2. Key takeaways
  3. Practical applications
  4. Recommendations