Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Emacs Headroom
Aug 2, 2003
Which is better:

1. Small consistent set of more in-depth interviews with standard rubrics, or
2. Large number of more conversational interviews with people you might work with?

I prefer the first, but I've noticed the second is picking up steam, especially for research-y groups (I think this is how MSR would interview; they probably got it from Bell Labs or something)

Adbot
ADBOT LOVES YOU

Emacs Headroom
Aug 2, 2003

DELETE CASCADE posted:

however, if you're a phd applying for a research job, there should be no rubric-based interviews anywhere. they should look at your publication record, then discuss how your research might fit in to their goals. it's just not possible to develop a standard rubric when hiring at this level. if you're just taking your phd to google for a software engineer level 3 job, expect rubrics, but not for a researcher

Assuming you're talking about a real reseach position (i.e. MSR / FB Research / Google Brain as opposed to some BI "data scientist"), how about :

"inside" research interview: interviewer is expected to read recent publications from the author and attend their job talk. interview will focus on:
  • how much of the research they developed on their own (vs. came from e.g. an advisor)
  • how impactful the research is
  • what caveats and limitations exist and what might be done to address them
  • whether the researcher has an idea of how to translate their results into products
  • (most importantly) whether the researcher has specific ideas to develop their own line of inquiry at the company

"outside" research interview: interviewer will present a new business problem, and the researcher will be expected to develop ideas for a line of research that can address it (e.g. "out sign-up page performs poorly for some groups" -> "oh let's try contextual bandits"). looking for
  • ability to think through an unfamiliar business problem
  • ability to come up with standard or novel scientific approaches to tackle the problem
  • good notion of ways to take the research and build it into systems
  • foresight in identifying low-hanging "v1" approaches, and how they can later be augmented

Emacs Headroom
Aug 2, 2003

qhat posted:

Maybe you're right, I wonder what an A-list company, say Amazon, pays an engineer in London. Let's have a look then:

SDE I: £38-49k, average £42k
SDE II: £44k-67k, average £56k
SSDE: £52k-97, average £81k

Welp. But sure it's "easy" to get £80k+ as a freaking noob in London because. Lol.

What are they issueing in RSU?

I know in the US Amazon has a company-wide salary cap of like $175k, with RSUs and bonuses (in proportion and timing designed to maximize retention) making up an increasingly large amount of comp as you climb the ladder.

Emacs Headroom
Aug 2, 2003

PleasureKevin posted:

this company has mandatory fields to asking you what % and $ amount you want to invest in the company. this is a job application.

it's a test, clearly

the correct answer is "zero: I act rationally and decouple my investments from my employment so that I minimize exposure to a single company or industry"

Emacs Headroom
Aug 2, 2003

qhat posted:

Maybe, but idk Scotland apart from everyone there has blood on their knuckles

and you'll have to take vitamin d supplements unless your skin is translucent

Emacs Headroom
Aug 2, 2003
coding test _after_ the onsite is unusual. how long did it take you? is this a super small start up who's nervous about making a hire?

i wouldn't ask for money, that would be super weird (and would have been something you negotiated -- like a "contract to hire" position -- before you agreed to do the coding test)

Emacs Headroom
Aug 2, 2003

PleasureKevin posted:

there was no onsite, just a Skype interview.

it took me probably 12 hours, but i'm gonna lie and say it took 6 to seem cool.

tell them how long it actually took; 12 hours is too long and they should shorten it in the future

edit: i've reviewed a shitload of take-homes in my day, and if one I assigned was on the too-long side, I'd appreciate the candid feedback (and the fact that you went ahead and did it anyways) more than I would be impressed by a candidate lowballing their self-reported time estimate. if you can get poo poo done and be honest that's pretty valuable

edit 2: if they're hostile to candidate feedback, or attempt to "explain" your feedback away, that's a pretty clear sign you might want to avoid working there

Emacs Headroom fucked around with this message at 21:39 on Jan 13, 2018

Emacs Headroom
Aug 2, 2003

PleasureKevin posted:

am I the only person getting actual work as "coding challenges"? 2 or 3 times they have been actually solving tickets of active projects.

who are these companies? name and shame

Emacs Headroom
Aug 2, 2003

qhat posted:

Not always. If they're being very unprofessional with you then chances are someone else will have the same complaint. Post at your own risk!

post away IMO

companies you're applying to aren't about to dig through glass door and try to de-anonymize posts; that's ridiculous. even outspoken drama-trolls like michael o church had to relentlessly make absurd and self-aggrandizing complaints about like 3 companies in a row over a span of years (getting banned from hacker news in the process) before he finally made himself famous enough to be radioactive

Emacs Headroom fucked around with this message at 02:01 on Jan 14, 2018

Emacs Headroom
Aug 2, 2003

maskenfreiheit posted:

i think you underestimate the ability of butthurt hr types at small to mid sized companies to get butthurt someone they can't currently retaliate against has increased the hurt in their butt

I promise that once you're gone they won't GAF. not only that but hr people are told by legal not to slander ex-employees

good recruiters and good managers do backchannel to each other (if they're in the same social network / same city), so if you're causing an actual problem like sexually harassing people, or punching a client or something, it could potentially get back to you. but they're really, really not going to dig through glass loving door to try to figure out who you are and retaliate.

ive seen people fired for really egregious stuff (like getting into fistfights at work) and come back with >5 offers a little while later after getting their poo poo together. as long as you're not shittalking your ex-employer during your interview, it's never gonna come up

Emacs Headroom
Aug 2, 2003

qhat posted:

You're some pretty bitter and delusional nerds if you think this is how normal people's brains work. Lol.

this times a million. nobody gives a poo poo about you or your snarky takedown of the product culture at pointless seed-round start up xyz on glassdoor

hr are more worried about being fired tomorrow since they're first to go if funding is an issue (maybe second if theres a sales or marketing team), or they're putting out some sexual harassment fire.

they look at glass door, but they're not gonna remember who you are 2 days after you leave, make that 2 minutes if you've been rejected for a job. unless you post something vitriolic with your name on it on medium, and it gets traction, they will have no clue who you are and won't care. anyone that does care (maybe obsessed founder?) is obviously so devoid of work or meaning that they are on a trajectory to obscure failure, so there's no reason to give them a second thought

if you don't want to write something up, don't, we don't give a poo poo either. but it's paranoid and delusional to think it's gonna get back to you

Emacs Headroom
Aug 2, 2003

jony neuemonic posted:

yeah i don't know why this is controversial. blasting someone on glassdoor probably isn't going to hurt you, but it definitely isn't going to help you so why bother?

because it might help other people deciding where to work? you don't have to be a total mercenary all the time

Emacs Headroom
Aug 2, 2003
good luck.

post your location / job title -- also how much is base vs. bonus vs stock? I'm at 260 base + 128k rsus in SF (just started new job)

Emacs Headroom
Aug 2, 2003

Notorious b.s.d. posted:

that's normal

moving to a high-CoL area never increases wages as much as the CoL

that wasn't true for me at least going from nyc -> sf

the sweet spot market is probably seattle. worst comp / CoL might be London

In the US, places like Chicago, Austin, and Boston have medium / lower comp but average CoL. places like Raleigh have lower comp but make up for it in very low CoL. NYC has medium-high CoL (but is the center of the universe, and the only functional american city with things like universal pre-K and a real transit system), and medium-high comp unless you're in finance.

SF has a high CoL and does not quality as a real functioning city, but the comp is the highest, and not just at FAANG companies

Emacs Headroom
Aug 2, 2003
that might be true.

there are other two advantages to being in big markets that I can think of:

1) if you can keep your costs of living the same as a proportion of your income, all things being equal it's better to be in a high comp / high CoL situation because while you're saving the same proportionately, you're saving more in total $$$

2) having a highly competitive market means you can easily move between jobs and aren't stuck in a trainwreck. I know a few people in philly who are tearing their hair out working for Comcast, but since it's more or less the only game in town, they're stuck

Emacs Headroom
Aug 2, 2003

ADINSX posted:

no joke, I have fig envy; I need to up my fig game soon, maybe this year.

A lot of it is really just bay area money. A friend of mine just did a pretty thorough search and got a spread of offers from 1.25M over 4 years to 2M over 4 years (some had significany amounys in ISO form which depends on liquidity events and performance though)

I didn't talk to Netflix but my understanding is that they're aggressive also with offers, but comp entirely with base, so like a $400k salary and that's it.

Emacs Headroom fucked around with this message at 01:47 on Jan 16, 2018

Emacs Headroom
Aug 2, 2003

qhat posted:

Hard cash salary IMO owns much more.

I agree, but a lot of places seem to cap at like 170 and go hard on stock

Emacs Headroom
Aug 2, 2003

FamDav posted:

first against the wall, etc.

i'm pretty far left politically (my goal would be "fully automated luxury space communism"), but since a lot of us are in uber capitalist america where the game is "try not to retire at 75 in debt eating catfood with your paltry social security benefits lol", you kinda have to shove money into a sack and be a little unapologetic about it i think.

esp. if you've (like me) spent most of your 20s in school not accruing any kind of retirement savings. and if you (like me) got one or more kids.

edit: its probably my socialist streak that compells me to tell people to post on glass door, and be radically transparent about salaries. everyone lets get more money from the VCs

Emacs Headroom fucked around with this message at 05:41 on Jan 16, 2018

Emacs Headroom
Aug 2, 2003

Schadenboner posted:

FTFY. Do not invest in individual equities, everyone at the table knows more than you do.

nobody knows anything and the S&P 500 will outperform most funds.

Emacs Headroom
Aug 2, 2003
i know someone who's doing a salary review in the bay area right now (data science / ML manager). here's what he told me about what they've found:

quote:

it's a broad range for DS, in bay area ive found it generally ranges between 300 and 600

there is a premium associated with ML / "AI" (now a buzzword) which pushes you into the higher end of that range

for someone with Ph.D. in technical field and 5 years industry experience with experience in ML, i would say median is low 400s

fresh ML phds start at least low-300s (and thats increasing rapidly) and we've seen some cases where a fresh phd gets offers as high as 500 (kind of ridiculous)

Emacs Headroom
Aug 2, 2003

qhat posted:

will be funny when everyone realises DL isn't all it's cut out to be and those salaries drop through the floor

much to my continued dismay and annoyance, DL is here to stay and works really, really well on quite a few problems

Emacs Headroom
Aug 2, 2003

Valeyard posted:

always seemed like amazon was the best way to go for foreigners wanting to work in usa tech

amazon has a comp structure designed almost entirely to lock you in -- big bonus and of first year, no stock until your cliff at year 2

thouhg if you're thinking of h1-b you'll be locked in anyplace, so...

Emacs Headroom
Aug 2, 2003

qhat posted:

Being an autist about their careers website is a red flag IMO. Lol.

even acknowledging the existence of the careeer website is probably a red flag

if you don't know someone there who can refer you (always the best method), then you're supposed to casually hit up a recruiter there on linked in

"hi, I'm making a career move and am in the middle of a job search. I've been a fan of netflix's technology culture [don't mention the product], and thought i should reach out to see what [general area like back-end or mobile or w/e] opportunities there are, and whether there might be a good fit. I'll be on the market for another couple of weeks, and am open to moving to the lower bay / SJ area if there's a match"

Emacs Headroom
Aug 2, 2003
I've been an a meeting with a recruiting team that went something like "uh, the resume backlog is kinda big... maybe we just clear anything past two weeks and shut down the req until head count gets figured out" more than once, at a very recognizable company

just sayin, if you go in the front door and never hear back, try knocking other places

Emacs Headroom
Aug 2, 2003

Kudaros posted:

Lots of midwestern universities top out at about 20-22k USD. I have a ~prestigious nasa fellowship~ and make 24k. For about 7 more days.

That salary review for San Fransisco is interesting... I'll soon have my PhD and definitely the background. Anyone recommend a way to get a salary investigation more comprehensive than glassdoor and the like for any given area?

have you done any internships yet?

Emacs Headroom
Aug 2, 2003

Kudaros posted:

Do you have any recommendations for internships? Why do you ask?

It's tougher to break in these days I think, with a large number of specialized ML PhDs coming on the scene (and Deep Mind / Google Brain / FAIR can't hire all of them).

I think an internship at a recognizable company is a really good idea if you want to move toward the ML / AI side of things. You can probably get hired as a "data scientist" (an "a-type" data scientist, that is) without an internship, but the role might be more skewed toward Business Intelligence (running SQL, making dashboards, analyzing A/B tests) and won't be as competitive on money.

Internship is also useful to benchmark the current market -- you'll be able to network with other people in your spot, and see what offers they're getting, etc.

edit: what is your field btw? climate stuff I'm guessing? fluid dynamics wouldn't be useful to most industries, but time series definitely would

Emacs Headroom fucked around with this message at 16:31 on Jan 23, 2018

Emacs Headroom
Aug 2, 2003
uh maybe call the cops? unidentified irritant in the bathroom is kinda scary

Emacs Headroom
Aug 2, 2003
i think i've only interviewed someone wearing a suit like 3 times. mostly people right out of school, or coming from a bank (even then most of the bank people cop onto the "just wear your normal-rear end clothes" standards in tech)

Emacs Headroom
Aug 2, 2003
i imagine its banks or crusty old monoliths like IBM that its normal to wear a suit?

not that I have anything against it (suits are comfortable, and nice suits are very stylish), ive just never even been inside a place like that in my short career in the tech biz

Emacs Headroom
Aug 2, 2003

jony neuemonic posted:

i'm expecting to gently caress up interview clothes a couple times before i get used to the east coast again. the whole west coast is so poorly dressed.

if its nyc, most people in tech dress noticeably worse than the rest of the city :P

the best-dressed people are the entry-level 20-somethings working in banks and marketing agencies

Emacs Headroom
Aug 2, 2003

Notorious b.s.d. posted:

ding ding ding

bloomberg? amex?

Emacs Headroom
Aug 2, 2003

The Management posted:

I can see where a still in college kid would think grades mean something. but they donít.

I think maybe banks will ask for transcripts, but they're totally ignored at any tech company I've been at (why would you care what grade someone got if a random person who learned everything on coursera outperforms them on the take-home and interviews?)

Emacs Headroom
Aug 2, 2003

Analytic Engine posted:

The Stanford label is the same as achieving and maintaining a high GPA: you (the interviewer) are expected to give more value to them (the Smart Person) because they are who they are.

ive interviewed dozens and dozens of candidates and screened hundreds of resumes across multiple recognizable tech companies. i can tell you i do not give a poo poo about college or gpa, and encourage anyone asking for advice in screening resumes to do the same.

projects matter, github kind-of matters, internships matter, significant OSS contributions to real projects (e.g. Apache-maintained) would more or less guarantee a hire

grades and highly-ranked college are good to have if you wanna go to business school.

edit: of course you have to be able to string a readable sentence together if you wanna go to business school, so maybe don't do that

Emacs Headroom
Aug 2, 2003

Boiled Water posted:

the order is view applications is likely resume -> github -> cover letter -> anything else -> grades

I, personally, have never read a cover letter. It takes me about 45 seconds to scan a resume, and the top things I'm looking for are experience (or projects), relevant pubs, and familiarity with frameworks / languages / tools the job uses

but I'm the type of interviewer that thinks whiteboarding some dynamic programming puzzles is a total waste of time.

Emacs Headroom
Aug 2, 2003

I HAVE GOUT posted:

How does an interview with you typically go? What questions do you ask? Do u spend most of the time going indepth on the points in their resume? What would indicate to you that the interviewee is someone to hire?

I like to set up a process that for junior people goes:
1) resume screen
2) take-home
3) phone screen
4) on-site

If I'm doing a phone screen, I'll typically try to talk about their take-home and ask them a few follow-up questions to make sure they actually did it and can think of potential improvements. I do ML and data science, so I'll ask about some of their previous projects (gauging how involved they were, how much they led vs were given a problem) and go into a few level-setting technical questions on basics of stats, regression problems, current ML methods, etc. If they haven't done much engineering, I'll also have them do a live coding exercise (typically something like streaming mean + reservoir sampling, or turning interaction timestamps into sessions).

For the on-site, one interview set-up I really think is useful is to have the candidate architect an entire ML system -- something like "let's send a targeted up-sell promotion to some users when they log in". Then have them pick a method (e.g. logistic regression), and we walk through making the necessary back-end systems, picking features, handling cold start, handling batch training / model serialization. It's really revealing if a candidate understands computers, they start thinking about what logging needs to exist and how they'll extract it; meanwhile others just talk about the algorithm and don't think at all about how actual computer systems can implement it in a way that it touches the product.

Emacs Headroom
Aug 2, 2003
I wouldn't say "write out the algorithm for reservoir sampling".

I'd say "let's say we have a stream of numbers showing up, they're floats and we don't know how big or small they are. I want you to write a function (or class if you want) that takes a single number, and returns the mean of all the numbers it has seen so far"

you'd write out a "naive" version (maybe after I tell you that closures or global variables in python are fine to use) that could have overflow issues, and then I'd guide to to refactor it so that it didn't anymore

Next I'd say "now let's do a similar problem. this time we have a really long list of numbers, we don't know how many, and we want to pull a single number at random from the list. can you do it without having to read the list twice to get the length?" and it should be somewhat clear what to do after the streaming mean problem.

In general though, if you want to do data science or ML, you should know enough engineering so that you can write working code and collaborate with engineers (e.g. "what's a microservice?" "what is deployment?" "what are code reviews?"). And you should be familiar with concepts like overfitting / cross validation, regularization, curse of dimensionality, frequentist hypothesis testing (maybe), bayes theorem and maybe bayesian inference, some unsupervised clustering methods, and some dimensionality reduction methods. I'd consider that the basic stuff, and things on top of that (statistical learning theory, bayesian hypothesis testing or multiple comparisons correction methods, fancy new methods like GANs or VAEs) to be nice but extra gravy. Some people are super deep in one particular area (like they only studied conv nets or something) and too shallow anywhere else to be useful.

Adbot
ADBOT LOVES YOU

Emacs Headroom
Aug 2, 2003

ultravoices posted:

i like how these threads turned from support group for job seekers to cranky bastards making GBS threads on anyone daring to seek employment in the noble profession of computer touching.

dunno if you're lumping me into the shitters, but at least from my pov I'm trying to give some perspective on what tech employers in places I've worked at care about so that people who want to work in tech have an idea of what to focus on

to someone who's busting their rear end getting perfect grades and doing nothing else at CMU or Stanford, it might sound disheartening to hear that those efforts don't matter so much and can still get your resume tossed in the trash (though investment banking woudl be delighted to take you). but to someone who did say arts for undergrad and is try to retrain themselves, or for someone who didn't have opportunities to attend a highly-ranked school, it's probably helpful for them to know that all they have to do is demonstrate they can do the work and they get hired

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply