Most DS jobs just need SQL and a bit of R. Don't overthink it.You're thinking entry level jobs for people with BA at most. That's not the job you're competing for with a PhD.
These guys have no idea. To get a PhD-level job since the pandemic, the standards have gone through the roof.
You need to be almost as good at python as a SWE, know SQL as well as a database engineer, and know stats as well as a recent stats grad at the masters level. And you have to know all of this on the spot.
Since interviews moved to virtual from on-site, it became much easier for people to interview. Recruiters and firms were overwhelmed by the number of applicants.
As a result, they raised the hiring bar significantly. Now, with the sharp downturn in 2022, standards have gone even higher these days. Not uncommon for a firm to collect 250 mid-career CVs and extend zero offers after the interview process.
Most DS jobs just need SQL and a bit of R. Don't overthink it.You're thinking entry level jobs for people with BA at most. That's not the job you're competing for with a PhD.
These guys have no idea. To get a PhD-level job since the pandemic, the standards have gone through the roof.
You need to be almost as good at python as a SWE, know SQL as well as a database engineer, and know stats as well as a recent stats grad at the masters level. And you have to know all of this on the spot.
in the end it is basically a pyramid scheme with all the bootcamps
in the end you can’t get a high paying job without the right connections, and many of these jobs are being axed bc of the recession since tech and finance companies can no longer pay for so many cushy sinecures
YOu cant get a good job without networking. Nobody trusts anything anymore. The resume, the degrees, etc. You need someone to tell someoen else "yeah this person is good".
Not true. You just need evidence that you’re good.
So if you are good, just build something that shows youre good. The recruiters and offers will come to you, you won’t even have to apply. They won’t even ask for references.
Works at a lower level too — if you’re not that great but can do the job, build something simple that shows you can do the job. This is one of the easiest careers to prove your worth in, free data is everywhere. Show you can do something useful with it, in an accessible way, and you will have no trouble finding a job.
(and your jmp is not useful or accessible, sorry. Build something your grandma could appreciate)
YOu cant get a good job without networking. Nobody trusts anything anymore. The resume, the degrees, etc. You need someone to tell someoen else "yeah this person is good".Not true. You just need evidence that you’re good.
So if you are good, just build something that shows youre good. The recruiters and offers will come to you, you won’t even have to apply. They won’t even ask for references.
Works at a lower level too — if you’re not that great but can do the job, build something simple that shows you can do the job. This is one of the easiest careers to prove your worth in, free data is everywhere. Show you can do something useful with it, in an accessible way, and you will have no trouble finding a job.
(and your jmp is not useful or accessible, sorry. Build something your grandma could appreciate)
The idea that a PhD thesis, especially from an HRM, in any way demonstrates less skill than most of the garbage being pushed out by the modal data scientist is plain wrong. Maybe instead of asking candidates to do dumber research, you should try to improve your ability to digest research as a hiring manager.
What do you even mean by "build" something? With what datasets and to what purpose? Most of what data scientists "build" in industry would get you laughed or kicked out of the PhD program. You're asking PhD candidates to forget economic and statistical rigor and just create some black-box, ad hoc predictive model that will wow nontechnical audiences. Maybe this is what's wrong with your industry in general and why the bubble bursting has been painful.
Or, you know, just keep hiring engineers to build fragile models with no robustness to slight changes in the environment, that sure seems to be holding up well in the current environment.
YOu cant get a good job without networking. Nobody trusts anything anymore. The resume, the degrees, etc. You need someone to tell someoen else "yeah this person is good".Not true. You just need evidence that you’re good.
So if you are good, just build something that shows youre good. The recruiters and offers will come to you, you won’t even have to apply. They won’t even ask for references.
Works at a lower level too — if you’re not that great but can do the job, build something simple that shows you can do the job. This is one of the easiest careers to prove your worth in, free data is everywhere. Show you can do something useful with it, in an accessible way, and you will have no trouble finding a job.
(and your jmp is not useful or accessible, sorry. Build something your grandma could appreciate)The idea that a PhD thesis, especially from an HRM, in any way demonstrates less skill than most of the garbage being pushed out by the modal data scientist is plain wrong. Maybe instead of asking candidates to do dumber research, you should try to improve your ability to digest research as a hiring manager.
What do you even mean by "build" something? With what datasets and to what purpose? Most of what data scientists "build" in industry would get you laughed or kicked out of the PhD program. You're asking PhD candidates to forget economic and statistical rigor and just create some black-box, ad hoc predictive model that will wow nontechnical audiences. Maybe this is what's wrong with your industry in general and why the bubble bursting has been painful.
Or, you know, just keep hiring engineers to build fragile models with no robustness to slight changes in the environment, that sure seems to be holding up well in the current environment.
I agree with you but you need a balance between rigor and being able to speak to a general audience. Remember that disaster of a Nobel press conference where Sims and Sargent couldn't offer any thoughts on the economy.
You're asking PhD candidates to forget economic and statistical rigor and just create some black-box, ad hoc predictive model that will wow nontechnical audiences. Maybe this is what's wrong with your industry in general and why the bubble bursting has been painful.
"economic and statistical rigor" = "I use diff-in-diff with doubly clustered SEs, because casual"
You're asking PhD candidates to forget economic and statistical rigor and just create some black-box, ad hoc predictive model that will wow nontechnical audiences. Maybe this is what's wrong with your industry in general and why the bubble bursting has been painful."economic and statistical rigor" = "I use diff-in-diff with doubly clustered SEs, because casual"
This
YOu cant get a good job without networking. Nobody trusts anything anymore. The resume, the degrees, etc. You need someone to tell someoen else "yeah this person is good".Not true. You just need evidence that you’re good.
So if you are good, just build something that shows youre good. The recruiters and offers will come to you, you won’t even have to apply. They won’t even ask for references.
Works at a lower level too — if you’re not that great but can do the job, build something simple that shows you can do the job. This is one of the easiest careers to prove your worth in, free data is everywhere. Show you can do something useful with it, in an accessible way, and you will have no trouble finding a job.
(and your jmp is not useful or accessible, sorry. Build something your grandma could appreciate)The idea that a PhD thesis, especially from an HRM, in any way demonstrates less skill than most of the garbage being pushed out by the modal data scientist is plain wrong. Maybe instead of asking candidates to do dumber research, you should try to improve your ability to digest research as a hiring manager.
What do you even mean by "build" something? With what datasets and to what purpose? Most of what data scientists "build" in industry would get you laughed or kicked out of the PhD program. You're asking PhD candidates to forget economic and statistical rigor and just create some black-box, ad hoc predictive model that will wow nontechnical audiences. Maybe this is what's wrong with your industry in general and why the bubble bursting has been painful.
Or, you know, just keep hiring engineers to build fragile models with no robustness to slight changes in the environment, that sure seems to be holding up well in the current environment.
There are PhDs who are clearly are very smart but who cannot code. I hired one, which was a mistake, and I've interviewed others. If the job involves coding, and you are interviewing for the job, and you haven't bothered to learn the well-publicized basics of the kinds of things we're going to ask you, that's a very bad sign. You assume people can just pick it up, but they can't necessarily, or they don't want to.
And when I look at code or ask coding questions, I'm not looking for any specific type of model. Your research would be fine. I'm looking for whether you coded it well and you can explain it.
Most data scientists are clueless. the last thing they want is some arrogant PhD stirring cheet up and making them feel doomb. We have a bunch of dudes in this thread who think putting some silly github project is good signaling, even better than a PhD. They're just not serious people anyways. The data science field is dying because of these clowns.
YOu cant get a good job without networking. Nobody trusts anything anymore. The resume, the degrees, etc. You need someone to tell someoen else "yeah this person is good".Not true. You just need evidence that you’re good.
So if you are good, just build something that shows youre good. The recruiters and offers will come to you, you won’t even have to apply. They won’t even ask for references.
Works at a lower level too — if you’re not that great but can do the job, build something simple that shows you can do the job. This is one of the easiest careers to prove your worth in, free data is everywhere. Show you can do something useful with it, in an accessible way, and you will have no trouble finding a job.
(and your jmp is not useful or accessible, sorry. Build something your grandma could appreciate)The idea that a PhD thesis, especially from an HRM, in any way demonstrates less skill than most of the garbage being pushed out by the modal data scientist is plain wrong. Maybe instead of asking candidates to do dumber research, you should try to improve your ability to digest research as a hiring manager.
What do you even mean by "build" something? With what datasets and to what purpose? Most of what data scientists "build" in industry would get you laughed or kicked out of the PhD program. You're asking PhD candidates to forget economic and statistical rigor and just create some black-box, ad hoc predictive model that will wow nontechnical audiences. Maybe this is what's wrong with your industry in general and why the bubble bursting has been painful.
Or, you know, just keep hiring engineers to build fragile models with no robustness to slight changes in the environment, that sure seems to be holding up well in the current environment.
PhDs are fine. I have one. I've worked with great data scientists with PhDs. But if I interview you and you have a PhD and you say you're a data scientist working in SQL and Python, but you can't answer extremely basic coding questions in those languages, we are not hiring you. And it's truly not because you're going to make us feel dumb.
Most data scientists are clueless. the last thing they want is some arrogant PhD stirring cheet up and making them feel doomb. We have a bunch of dudes in this thread who think putting some silly github project is good signaling, even better than a PhD. They're just not serious people anyways. The data science field is dying because of these clowns.
YOu cant get a good job without networking. Nobody trusts anything anymore. The resume, the degrees, etc. You need someone to tell someoen else "yeah this person is good".
Not true. You just need evidence that you’re good.
So if you are good, just build something that shows youre good. The recruiters and offers will come to you, you won’t even have to apply. They won’t even ask for references.
Works at a lower level too — if you’re not that great but can do the job, build something simple that shows you can do the job. This is one of the easiest careers to prove your worth in, free data is everywhere. Show you can do something useful with it, in an accessible way, and you will have no trouble finding a job.
(and your jmp is not useful or accessible, sorry. Build something your grandma could appreciate)The idea that a PhD thesis, especially from an HRM, in any way demonstrates less skill than most of the garbage being pushed out by the modal data scientist is plain wrong. Maybe instead of asking candidates to do dumber research, you should try to improve your ability to digest research as a hiring manager.
What do you even mean by "build" something? With what datasets and to what purpose? Most of what data scientists "build" in industry would get you laughed or kicked out of the PhD program. You're asking PhD candidates to forget economic and statistical rigor and just create some black-box, ad hoc predictive model that will wow nontechnical audiences. Maybe this is what's wrong with your industry in general and why the bubble bursting has been painful.
Or, you know, just keep hiring engineers to build fragile models with no robustness to slight changes in the environment, that sure seems to be holding up well in the current environment.
The thing is, if you're smart enough to get a Phd, you don't get a Phd. There are many data scientists with just a BA or BS who are much brighter than Phd grads. Being good at theory or mathurbation does not translate to effective industry data science.
Since interviews moved to virtual from on-site, it became much easier for people to interview. Recruiters and firms were overwhelmed by the number of applicants.
As a result, they raised the hiring bar significantly. Now, with the sharp downturn in 2022, standards have gone even higher these days. Not uncommon for a firm to collect 250 mid-career CVs and extend zero offers after the interview process.
I'm at a vlrm university. The last time we had an open field search, we got 600 applications. Granted, some were from HRM that were not applying seriously, but you're kidding yourself if you think 250 applicants for one job is going to impress anyone on the academic buy side. There are decent Nonademic jobs for Phds,, but your applicant numbers suggest you don't have one of them.