webspiderus

yep

how do you feel about still having to use outlook because of concerns about a competitor that's not even doing social anymore? I've read FB SWEs be aghast about people uploading whiteboard photos to Google Photos to be searchable later, paranoid about Larry and Sergey poring over all the data while twirling their evil mustache.

Also, if impressing other people is not what you're optimizing for then you might not have a great trajectory at a company that's obsessed with i m p a c t and hallucinating metrics to measure it.

Some parts of working at FB as a SWE are awesome, some aren't, everyone's mileage certainly varies.

really, Picasa led to Google Photos as the previous photo product the company had ..

some of that team helped build the Google Photos app, but Google Photos was a product in its own right even before Bump joined the company and they were part of a (much) larger effort, of course.

The research fix may have been to train their data better but the blacklist of bad terms was as real of a product fix as it gets.

Well, to be fair they excluded high tens / low hundreds of potentially offensive terms from search before even launching and when this came out they just extended the list a little. Sometimes having product vision requires recognizing that the products you build come with limitations and potential for very real emotional reactions of very real human being users.

Sure, you can create a set of 'face templates' that may or not be linked to a profile, and then match against them as long as you have some meaningful way to use the matching templates without a profile.

unfortunately, it certainly doesn't seem like the company does this all the time. i got placed onto a team which has no bearing on my background experience whatsoever, and after approaching my tech lead and some HR people to continue the conversation, everyone else decided to agree that the 18 months guideline was "fair" and that the 20% projects were there to satisfy my need for interesting problems.

i had potential teams lined up (one of which I would've been allocated to initially, had they not hit their new grad hire cap), so it was certainly a very frustrating experience. perhaps they care more about established engineers than recent graduates?

that's weird, because my experience is exactly the opposite! I've spent about 4 months on my team so far, which has very little to do with my background, and after I realized just how bad the fit was, I started a conversation with my tech lead, his supervisors, and some HR people about moving to a team that would be a better fit (which had openings at the time).

3 weeks of conversations later, I was told that the 18 month rule was "fair" in everyone's opinion, and that if I'm really concerned about working on projects that are interesting to me, I should just find one to spend my 20% time on (easier said than done).

This is also not entirely accurate. I got my offer a good 5 months before I even knew which teams I had a chance to speak with before the allocation process, let alone before I knew which team I would be placed on.

I think this should also have the added effect of being overall faster, as doing lots of addition would be quicker than doing lots of multiplication (log notwithstanding).

As another sample, I'll say I'm far from feeling swindled. I was a bit disappointed that the publicly available ML class and its Stanford equivalent only differed in one optional discussion a week - but I certainly don't feel the time taken developing the class took away from the in person experience, and I do think Andrew Ng and the staff were more available for Stanford students than they were for those taking the class online (which I view as the justification for the Stanford tuition costs).

The cryptography class I'm taking this quarter is more to my liking in that aspect, as in addition to the video lectures (which I assume will make up the bulk of the public offering), there is another 2 hours of lecture and a discussion that cover additional material. In this kind of a set up, the added benefit for the Stanford students from an educational standpoint certainly seems more tangible.

it's interesting because I just got the elevator question for the first time a month or two ago and was still scratching my head trying to figure out where the decision point on that question is. i ended up spending about half the interview talking about different ways to optimize for minimal waiting time, but now it seems like that may have just been banter after we'd covered the FIFO queue.

completely agree. as someone that's just about to graduate, i have to say i was a little shocked when i figured out this is the way things are. the difference between the first engineer and one of the founders never seemed so categorical to me, and this was one of the few things that made me feel it's not really worth joining a startup as early as I'd like (to work on interesting+useful things, have impact, etc.) unless I'm actually involved at the founding stage.

I wonder if it's the link in "All content printed on 100% recycled internet memes"?

In my high school public speaking class, we also had a goal to remove these stop words from our speech by the end of the semester - but in our case, the teacher would simply stop you if you ever said one of these stop words in class, at which point you would have to say what you were going to say again without any stop words.

it didn't take long for most people to learn, and it's probably one of the more valuable lessons I've carried with me from high school. I've learned to be more comfortable taking a pause if I need it, which I think has had only positive effects.

On the other hand, sometimes it can be annoying just how aware you are of other people using stop words :)

Absolutely (in the US here). Some carriers will allow you to block all text messages, which can get rid of the annoying ones you wouldn't want to accept, but also forces legitimate ones to get eaten - causing some people to think you're ignoring them when you actually had no idea they even tried texting you in the first place.

also often used for school mascots, at places such as Duke, my high school, and many others surely.

I've had a similar gripe with scrolling, although I encountered with text editors instead of browsers. Having used Sublime Text for a while (which does what the article suggests and shows you as much white space as is needed to keep the display consistent), it is really jarring using many of the other editors which refuse to do this, especially as I often tend to write new code at the bottom of the page.

Being forced to consistently look at the bottom of my screen instead of the top or the middle is pretty annoying, maybe a change in browsers could trigger a change in text editors as well!

F.lux is actually available for all platforms! I've been using it on Windows for a while, one of my favorite small utilities.

this is literally the first thing taught in both of the machine learning classes I've taken from Prof. Ng at Stanford, so maybe it's the application of logistic regression more than the estimation technique itself that makes machine learning?

for what it's worth, I'm taking Stanford's version of the ML class that Prof. Ng is offering, and aside from being a couple of weeks and ahead and having to do a final project, it is pretty much the same offering as that available the outside world. I think this could explain the level of quality of the website, or at the very least it having more of a course feeling to it.

for what it's worth, I'm taking Stanford's version of the ML class that Prof. Ng is offering, and aside from being a couple of weeks and ahead and having to do a final project, it is pretty much the same offering as that available the outside world.

To speak for the latter, Google's 10% across-the-board raise last year does apply to new hires, personally speaking. I'm surprised that my offer is above the MS median and average.

this is the the same class. having taken it last year (and taking Stanford's version of ml-class), this is a great choice for a more thorough exploration of the (mathematically) finer points of the methods covered in ml-class.

Although I'm hoping to stop doing that soon, I have been able to avoid watching many of the videos so far since they seem to be very similar to the concepts covered in the class last year. Prof. Ng's notes for this class (available at http://cs229.stanford.edu/materials.html) are also quite helpful for refreshing the finer points of algorithms.

so, is ml-class.com then more comparable to the CS 229A class that is being offered this year? I'm taking the latter, and was wondering how similar the two are.

having taken the 224W, 229, and 246 courses, I will say that they provide a good introduction to a lot of the data mining algorithms (with 224w emphasizing graph-based algorithms, 246 emphasizing things like decision trees and association rules, and 229 emphasizing regression and SVMs).

however, the main issue for me is bridging the gap between completing those courses and actually being able to apply the skills learned to real-world problems. i hoped to do that with my internship this past summer, but it proved to be more difficult than i thought, particularly without having much guidance from anyone with similar specializations.

how much value does a certificate like this signal then? i feel like there's not much it shows beyond commitment and a focused interest, since it seems to me that the true test of anyone in data mining comes in the form of projects, not classes.

just looks like "this" needs to be properly disambiguated :)

I definitely agree with your last point - part of the reason why I decided to take a machine learning class again from Prof. Ng was because I still had a sense of unease that I wasn't quite getting the topics that would really be important when I'd implement machine learning algorithms (whatever they end up being).

Mostly though, I'm just hoping I can find a place to work with plenty of ML people to learn from :)

the Machine Learning class (at least for the Stanford students) also has a project component, so I'm sure a startup could see some use even from just having the graduates work on their problems for a research project, if nothing else! personally, I ended up doing two projects last year with data from my previous internship, and I think it's a great way to both have interesting problems and useful results.