Shreya Shankar
PhD student in databases focusing on MLOps @Berkeley_EECS @UCBEPIC. Prev. 1st ML engineer @viaduct_ai, research @googlebrain, BS & MS @Stanford CS. She/they.
View on πThreads
I enjoy reading thoughtful posts about career growth & optimization. This one made me think for a couple of days. I'd like to offer an alternative view -- that MLOps is quite techn...
Adding to this point: "the algorithm" means different things to everyone. Here is a snippet from my predictive modeling retro I wrote in 2020: https://t.co/su2tIt9G6Q https://t.co/...
I really like this MLOpsy research vision (h/t to @AlexTamkin for sharing with me a while back) so I will take the opportunity to go on my tangential versioning rant: while version...
With each day I feel like I know less about modern-day ML/DL π but I'm excited to talk about what I think is the biggest failure mode in many low-code (eg AutoML) systems: they are...
This is a nice thread. I similarly have been admiring recent AI demos but struggle to articulate why I'm not in an excited frenzy. I think it's bc the results are precisely what I'...
Agreed. One interesting nuance we are noticing in our research study: when people think, "how do I apply unit testing / DevOps principle X to my ML pipeline for the sake of having...
Can't stop thinking about a recent discussion with a bigco ML eng about how data scientists/MLEs tend to optimize for "keeping their GPUs warm" (i.e., running all sorts of expts to...
Ok y'all, grad school has taught me 2 things: (1) put the *correct* date on your calendar and (2) read the fine print. I need to get my NSF GRFP story off my chest (please ignore i...
Seeing all these threads about habits of good ML modelers on this app but many of them go out the window when it comes to prod ML. For instance, it's possible to have "too much" ex...
I love this site. I will randomly be scrolling through my feed and see something insanely cool like this, watch the video (https://t.co/De3YJlPvUP), and get very excited because ML...
I love this post: the infra team at a company was drowning in devops requests so they built an end-to-end platform for devs. Devs don't want to use it. Why?? IMO in these cases:...
I'm currently procrastinating on writing so I will write a thread on writing. This is mainly for STEM people who keep saying they want to write more. 1/9