← sources

How can I really "Git Gud"?

source: reddit (reddit:datascience)published 2026-10-28 17:20ingested 2026-10-29 02:38

retained text

I have been giving interviews over the last few months and I have categorized most of the hits and misses into various buckets I need to work on.

One of them is scenario based "why and why not" style questions? Why would a model work, why not so and so method etc. My problem is I am not really good at this - I can explain the decisions I made in my projects, but if the interviewer tweaks something I will likely not be able to answer this.

>I have also been thinking in gaming terms (I am an avid gamer) and a passionate DoTA and R6 Siege player. I have, with a little shame, put \~3K hours each in both. But that grants me a vast knowledge (*not an expert*) where I can play with scenarios and theorycraft stuff and matchmaking keeps pitching me new challenges.

Akin to this, how do I get this level? Working on projects isn't helping a lot, as I am sticking to stuff I know. Since *I don't know what I don't know*, it's really hard to grow. 

document fetched 2026-10-29 00:25 · canonical https://reddit.com/r/datascience/comments/1xhucmt/how_can_i_really_git_gud