Get live statistics and analysis of Gowthami's profile on X / Twitter

Growing up in small town in India, you can see greatness from far, someone reached moon, nasa put a rover on mars… you can see the greatness but you are not sure you can be more…. We moved to US as an experiment… masters became PhD… and then my partner joined Optimus… suddenly you felt like you can touch the greatness and you can be part of it… That’s what US does to you, you can be nobody one moment and next moment you can be part of history! Even though you aren’t conventional candidate , they take chances on you. I’m sure no Indian university would’ve taken a chance on me (trust me I tried), but UMD did. That's the true power of this place. It doesn't just change your opportunities; it fundamentally changes your own perception of what you are capable of!

For ML roles, instead of asking candidates to solve a LeetCode problem, ask them to list a few prompts from a benchmark relevant to their projects. This reveals if they’re truly battle-tested or just bs-ing!

A piece of unsolicited advice. I see a lot of internship calls these days from many companies on X. Usually one chooses an offer based on perceived brand if you have multiple offers. But what I’m noticing lately is, companies are using internships to continue their publication pipelines since it’s cheaper and still maintain their “research brand”. There are a lot more internships than actual RS roles in these companies. So if you are in final year, try to optimize for company that actually gives you a job later, not someone that’ll make you interview from scratch along with other candidates. Another thing I also noticed in academic communities is, somehow internships in applied teams are looked down upon than fundamental teams. This mindset is so strange for me cuz I worked in industry before grad school and applied research is bread and butter and “fundamentals research” is extra fat which gets cut first. So yeah, if you’re optimizing for industry career, look for applied internships cuz usually those are the teams with headcounts later. Most of this is quite anecdotal, please use it with caution. 🙂

Oh, forgot to mention - you are more likely to get internship interviews or NeurIPS party invites if you call yourself final year. 😉

Any good resources for understanding "Norms" in transformer models? They don't make sense to me at all! How can a LayerNorm vs RMSNorm make so much difference? How can one hypothesize which norm is the best for the model? Drop resources in the comments, please. 🙏

Here’s my take on why coding interviews are a lossy approximation of a candidate’s ability. Most of time, a candidate gets rejected cuz they aren’t able to finish the question, however real-world coding is not time bound. You can be an excellent programmer but how well you do in an interview depends on how in-distribution the interview question is, or how comfortable you are with the interview format. So the system is rigged to reward people who are good at interviewing rather than someone who can solve a (perhaps) an out of distribution problem given enough time. Another thing I noticed is, given how good coding LLMs are… you should be testing the general understanding, like why you do certain thing or how do you eval something rather than how well you can rote-memorize Attention forward and backward passes. 🤷‍♀️

Growing up in small town in India, you can see greatness from far, someone reached moon, nasa put a rover on mars… you can see the greatness but you are not sure you can be more…. We moved to US as an experiment… masters became PhD… and then my partner joined Optimus… suddenly you felt like you can touch the greatness and you can be part of it… That’s what US does to you, you can be nobody one moment and next moment you can be part of history! Even though you aren’t conventional candidate , they take chances on you. I’m sure no Indian university would’ve taken a chance on me (trust me I tried), but UMD did. That's the true power of this place. It doesn't just change your opportunities; it fundamentally changes your own perception of what you are capable of!

Any good resources for understanding "Norms" in transformer models? They don't make sense to me at all! How can a LayerNorm vs RMSNorm make so much difference? How can one hypothesize which norm is the best for the model? Drop resources in the comments, please. 🙏

For ML roles, instead of asking candidates to solve a LeetCode problem, ask them to list a few prompts from a benchmark relevant to their projects. This reveals if they’re truly battle-tested or just bs-ing!

> be me > come across a paper with interesting premise. > excitedly start reading > claims are on mnist/cifar > all excitement gone, reduced to atoms

Here’s my take on why coding interviews are a lossy approximation of a candidate’s ability. Most of time, a candidate gets rejected cuz they aren’t able to finish the question, however real-world coding is not time bound. You can be an excellent programmer but how well you do in an interview depends on how in-distribution the interview question is, or how comfortable you are with the interview format. So the system is rigged to reward people who are good at interviewing rather than someone who can solve a (perhaps) an out of distribution problem given enough time. Another thing I noticed is, given how good coding LLMs are… you should be testing the general understanding, like why you do certain thing or how do you eval something rather than how well you can rote-memorize Attention forward and backward passes. 🤷‍♀️