Resources

S. Keshav's classic three-pass method for effectively reading papers.

AI-assisted literature search with citation graphs and one-line summaries.

Upload papers and query them, or generate a podcast to digest them on the go.

Bill Freeman's notes on what gets a paper accepted. Equally valid for ML and beyond.

Vincent Lepetit's comprehensive slide deck on what makes a paper read well.

Michael Black's «the secrets I share with my students.»

Jacob Steinhardt on the mechanics: sentence- and word-level advice.

Devi Parikh on the writing process. This has shaped how I approach paper writing.

Kate Saenko's system to nail the introduction. This is the method I use.

Strunk's timeless writing guide. The free Gutenberg copy is the 1918 original; the modern Strunk & White revision is the standard reference.

Edward Tufte's foundational book on showing data honestly and clearly. You'll never unsee a bad graphic again.

Markus Püschel's practical tutorial on tables that are easy to read.

Free diagram editor for figures and system sketches.

Strips comments, unused files, and clutter before you upload to arXiv.

Shrink an oversized PDF from the command line (gs -dPDFSETTINGS=/prepress …). Prefer vector graphics first.

Simon Peyton Jones' guide to giving a talk that lands. I rewatch it before my own talks.

The #BetterPoster approach: one big takeaway instead of a wall of text. A bit extreme as a template, but it nails the priority.

Edward Tufte's critique of slide-think — how bullet-point decks degrade reasoning.

Devi Parikh's group on writing rebuttals that actually change a reviewer's mind.

CVPR tutorial (video + slides) on being the reviewer you wish you had.

Countdown to the major ML/AI submission deadlines.

The same, for the top security and privacy venues.

ETH's searchable index of CS conferences and their dates.

Ranks institutions and researchers by publications at top venues. Handy when deciding where to apply.

MIT's crash course in the tools every researcher should know: shell, git, debuggers.

Runtime type checking for Python. Catches subtle errors plain annotations won't.

Shape- and dtype-annotate tensors (PyTorch, JAX, NumPy). Documentation and runtime checks in one.