|
I recently went through a bunch of interviews, so if anyone is curious about questions you might hear at facebook or random startups, feel free to ask.Vegetable posted:Is there a need to learn ETL and data engineering stuff? Because I’m a data analyst and all I do is write queries to pull data. Wondering if I need to extend my SQL skills if I do make the move into data science. Not really although occasionally a startup wants a data scientist who will also be their data engineer. Suspicious Lump posted:So probably a silly question but apart from extracting data from a server, is SQL used for anything else? I'm trying to understand why SQL skills are so sort after. Yeah, absolutely. You can do some basic analysis and modeling in SQL which comes in handy when you have giant data sets. If you want to run a linear model for some forecast daily for 10,000 customers, it's nice to do that in database. It's particularly nice when you have a database which supports user defined functions so you can apply some model in Python or R across the db. pokie fucked around with this message at 08:01 on Jan 21, 2019 |
# ¿ Jan 21, 2019 07:58 |
|
|
# ¿ May 22, 2024 12:20 |
|
BattleMoose posted:So when applying for data science jobs, what would be the best way to communicate a portfolio of coding/data jobs you have worked on? Is having a personal website the best way to do this? Attach it to your resume when submitting? I just have a resume that's basically a short skills/educ summary and a list of projects a few of which have more elaborate descriptions. I used to try to fit it all on one page, but recruiters told me not to. I have been in the field for about 6 years fwiw.
|
# ¿ Jan 23, 2019 19:40 |
|
MickeyFinn posted:My brother forwarded me an invite he got to some data science boot camp called The Data Incubator, I chucked them a resume and had a few friends write references because applying for jobs is a numbers/affinity game, so why not? Big surprise, I'm now a "semi-finalist" which gives me the "opportunity" to spend the weekend, already booked with 30-something hours of work, doing their challenge puzzles in the hope that I will tick all the right boxes. The best part is the so-called project that I need to propose to work on while I'm there at the 6-week "program." Because... Your take sounds correct to me. From the second hand experience I have of data boot camps they function by pre-screening their candidates so much that the finalists are all but guaranteed to succeed. A guy I knew went through one. He is a physics PhD with multiple publications who worked on one of the colliders. I mean, c'mon.
|
# ¿ Jan 27, 2019 20:37 |
|
MickeyFinn posted:That’s me, too! Except for working on colliders, I build them. Suspicious Lump posted:For me, every boot camp I come across always seems to focus on the wrong aspects. For example, data science bootcamps IMO should be stats heavy, programming light but they always seem to flip it. I think that's because so much of the typical interview process is focused on fizzbuzz, data structure, sorting algorithm nonsense (and yes, I realize a lot of that is interesting and useful but imo not as a way to judge data science candidates). pokie fucked around with this message at 22:38 on Jan 28, 2019 |
# ¿ Jan 28, 2019 20:51 |
|
BattleMoose posted:I have a stem PhD and did a post-doc, for about 6 years of my life my workload was probably about 70% coding (python and ncl) and 30% science and other academic related activities, I also tutor statistics. The biggest project I was working on was using data from 25 global climate models to predict changes, in total about 2000 years of simulation were used. This project was 90% coding and largely data management, looked at multivariable linear regressions but in the end a simple linear regression worked, albeit an enormous amount of data was used. I have an MS in stats (after leaving a PhD program) and it took me a year to find my first job. I did that by going to relevant professional events, like R and python meetups.
|
# ¿ Feb 2, 2019 09:36 |
|
MickeyFinn posted:It's probably a conspiracy theory to say they stole the idea from him, but given Facebook's lovely history it wouldn't surprise me. It is far more amusing to me that an idea declared stupid in an interview was used literally two weeks later to do the thing it was pitched to do. Anything is possible. One of the recruiters I work with is involved in a law suit where a company they referred a candidate to brainfucked him and built a product from the interview, apparently. However the example you give is straightforward enough that I find it hard to believe the 100+ data scientists at fb couldn't come up with it on their own.
|
# ¿ Feb 5, 2019 18:40 |
|
MickeyFinn posted:
If you are in CA, I recommend going to R and possibly other related meetups. I got my first full time job this way fwiw. Also look at angel list job postings - it's easier to get started at a crazy startup.
|
# ¿ Mar 6, 2019 19:50 |
|
meanolmrcloud posted:Last few weeks of my bootcamp are pretty intense. I’m doing a few phone interviews a week, and finically getting the hang of answering technical questions with confidence but I should probably keep reviewing notes and doing practice problems. We were also slotted into our groups for the final project, which is given as a demo in front of all the tech companies in town. The only two people booted out of the camp for failing the final exam were in my group, so it’s just me and another person, which is probably a good thing. All of this while doing my full time regular job means it’s pretty stressful. Hang in there, buddy. I hope it works out for you. Out of curiosity - did they give you any useful interview prep?
|
# ¿ Mar 6, 2019 21:03 |
|
meanolmrcloud posted:Yea, we had a few rounds of mock-formal interviews with some of the directors, who provided jobs and roles at different types of companies to practice aiming for their level of casualness/professionalism. We also had some technical interview prep, where we were given direct feedback and suggestions. Still, I’m very used to soft-science interviews, and it’s stressful to have to recall and nail technical details. Cool. Some of it gets easier - I feel like I have nailed every interview about model creation so far, but a lot of it can be specific little bullshit. It's easy to be anxious about it. I've been at it for 6 years and still get pretty nervous at interviews. Some guy was grilling me on Simpson's paradox today, which is not something I have touched since undergrad.
|
# ¿ Mar 8, 2019 06:22 |
|
TBH this sounds like a terrible idea. From my understanding the value that boot camps bring is less in what they teach you and more in the employer network they have. If the camp is not local, what use is their network to you? I suppose you could uproot and move etc.
|
# ¿ Jul 19, 2019 03:26 |
|
|
# ¿ May 22, 2024 12:20 |
|
JIZZ DENOUEMENT posted:When I was in grad school, I really liked my courses on program evaluation and regression analysis. Stuff like STATA. They certainly exist for stuff like SAS, but unless you want to work with really old school businesses, you would want to use Python and R. I am not aware of certificates for either that anyone cares about. There is a crapton of resources on data science programming online. You could learn on your own and participate in competitions like Kaggle. Post your code online and show that to employers.
|
# ¿ Oct 26, 2019 07:16 |