How I Work

Here is Cal Newport’s Golden Rule of productivity says: Deep work produced = time working * intensity of work

Most time is not spent writing new code

Most time is spent debugging stuff Most time is spent re-running experiments because you messed something up before Optimize to avoid these mistakes Check outputs frequently

Setup and productivity

Block off time on your calendar for each task Only listen to music if needed, preferably music without lyrics Make sure you have water nearby Log deep work hours on a calendar next to your desk Don’t check email or get distracted by minor things


When you’re running lots of experiments, use IDs to label do import utils instead of from utils import train Using aliases

Biggest time-wasters

Re-reading old email or slack conversations Reading random blog posts online Looking at online profiles/twitter accounts of other researchers Looking at previous stuff that I wrote because it has cool formatting or I like the sound of my own writing Applying to jobs/positions that you aren’t super interested in Close tabs that you aren’t looking at Admiring your LaTeX stuff

Principles of research

Creativity First-principles thinking Intensity Simplicity Honesty Depth of analysis Long-term vision: Play the long game