Reading, writing & talks
The skills that matter as much as the code. Great research that no one can follow doesn't travel — learn to read efficiently, write clearly, and present well.
Reading & writing
- Read papers actively — for any paper, write three sentences: what problem, what idea, what result. That habit is worth more than reading twice as many. Skim the abstract and intro first; go deep only where it matters.
- Write as you go — a report or thesis grows as you work, so by the end you're revising, not scrambling. Keep it concise and grounded: claims tied to evidence, no filler.
- Learn a bit of LaTeX — Overleaf is browser-based LaTeX and the standard for research writing. Its "Learn LaTeX in 30 minutes" is enough to get going.
- Read the masters on it — Bill Freeman's (MIT) chapter How to Write Papers, in the free Foundations of Computer Vision, is short and full of concrete advice.
Giving talks
Presenting your work well is a skill you can learn — and these are the standard references for it. All three are short (Winston's is an hour well spent once).
Simon Peyton Jones: How to Give a Great Research Talk →
Michael Ernst: How to Give a Technical Presentation →
Patrick Winston: How to Speak (MIT) →
Bill Freeman: How to Give Talks (free chapter) →
- Your talk is an advertisement for the work, not the full paper — give intuition, omit the gory details.
- Lead with the problem and why it matters, ideally with a concrete example. Don't open on a title slide you dwell on.
- One idea per slide — descriptive slide titles, few words, real figures.
- Practice out loud at least three times, including once in front of a live person. Nerves are universal and fine.
That's the guide. If anything here is unclear or you want a pointer on where to go next, feel free to reach out — and good luck.