boomzilla

That's interesting. What is the input into the process? Don't you need a PRD or a requirement doc to start with?

How about this guide:

https://docs.nvidia.com/cuda/cuda-programming-guide/pdf/cuda...

sorry could not resist the pun

Two words: plausible deniability

how would failover work?

Peter's principle and whatnot, but I think there is something deeper. The manager positions are designed, by definition of the word, to manage, and the top goal is to extract values from workers. Managers (and product managers in tech companies) are encouraged to create a `healthy` tension with line workers (software engineers included) in work estimation and commitments. This is supposed to make the work challenging enough, but not so demanding that burn out the workers. The best managers can do that by providing the intellectual challenges and motivational goals. Most resort to processes and plain OKRs though (reflected in the worst ever software tool, JIRA. Also any line manager who's got any clue, meaning who can provide technical/business directions, would be quickly promoted to directors (where they are supposed to direct :-).

Protip for frontline managers: The percentage of time you spend on JIRA is negatively correlated to the chance of being promoted to the director level.

I am of the opinion that `true AI` is the science/engineering of understanding and replicating human intelligence. Why are we able to come up with abstract concepts from the surrounding physical environments? Why do we look at the stars and wonder what they are (and why)? How are we able to communicate with one another through pictures, words, writings, snapchat. Is that something special about our brains, our collective society, or something else, that enables such remarkable different behaviors from other any animal on earth? I don't know which direction we can start to go down to answer these questions, but collecting good data sets is probably as good as anything. Maybe we'll get the `quantity` of smarter specialized systems first, and once we get the `quantity`, maybe the `quality` will follow?

Well then the OP ignores the biggest advertiser of them all: the communist party.

The AirBnB story about `literally a month from being homeless` is total BS. Both Chesky and Gebbia worked for a few years before starting the company and the other guy went to Harvard.

You can read the comments (and the linked papers) first. This is an advanced algorithm that could take days (or weeks) to fully internalize the details. One can't expect to just read the code and build a mental model of the program in one parse, no matter how expressive the variable names are.

Don't think there is much rationale behind all these models. It's more like P(would buy a vacuum | bought a vacuum) > P(would buy X | bought a vacuum) where X is a single product. Now P(would buy a vacuum | bought a vacuum) < sum(P (would buy X | bought a vacuum)) for X that is not a vacuum, but what would be the recommendation? Hey, you bought a vacuum, come back and buy some non-vacuum stuff?

For most recommendation UIs, you would need a hero item that make people want to click on. It might turn out that another vacuum is probably the best item for some people to click on, and go on to buy other stuff once they are on the site.

Be frugal. Don't buy a house unless you have at least 6 months of payments in the bank. Don't buy a car if you need to take a loan. Save as much you can. Build and maintain a strong network outside work (family/friends/professional contacts).

Every time I got into a difficult situation at work, I take a deep breath and tell myself: "Don't worry, give it your best shot to resolve this. And if that's not good enough, you know you can walk out that door and take a break for some time". It's been working well for me.

This is a better link: http://www.businessinsider.in/an-ex-military-contractor-work...

how would `an iOS app for listening to podcasts focused on social discovery` make money? I couldn't find that question in the interview.

No the idea is that you invest on the real estate with your own money, and invest on start up with LPs' money.

The original word2vec, written in C, could utilize all the cores available. It's actually refreshing to read that code, a true one person shop, hard engineering code :)

So are logic frameworks discovered or invented? Are true/false values discovered or invented?

I think the question whether math is discovered or invented is silly. It is both.

That's very childish and lame for Steve Huffman to say "Ellen wasn't the first Reddit engineer, so she probably lacked the expertise to do it, and even if she did, she was smart enough to not".

I think this guy needs to grow up.

That's exactly the point. No one has ever offered such a document, so say what you want about Trump, he's taking a step towards transparency and accountability.

Er, should GAAP net income include SBC? Otherwise, it's non GAAP.

And how will they make money? Selling ads in the AR world?

It's kind of a different software that are being developed these days. Software development used to be highly original, analytical and creative, but now it's like endlessly iterative on the same problems and tools. That explains the number of programmers/engineers around. My guess is that the highly selective group of engineers who are still working on the most original problems sit in their own office (preferably their home office, e.g. Linus). For the rest of us, it's more like a factory floor.

I think you misunderstand the purposes of auditing. They have very strict accounting that they have to follow. There is a degree of "interpretation" on what numbers can be assigned to certain items, but by and large, they just follow a big book of rules. Of course there are certain new business practices that are not reflected fairly and properly by these rules, but that should be the exception, not the norm.

A few takeaways:

- Morgan owns a small percentage of Palantir, maybe 20 basic points. However, this is big as Morgan and the big i-banks still have a lot of influence on the big money funds. This will scare the LPs off and VC funding will dry up.

- In my opinion, Palantir is still too expensive at ~15B valuation. I mean how can it be more expensive than both Tableau, Splunk and Hortonworks combined in the public market?

- Dropbox seems to be hit the hardest when T. Rowe Price marked it down to just 5B valuation. Still a great outcome if they can exit at the price point, but it sucks for the late investors and recent employees.

I think you are biased to pick a "micro framework" as a starting point. Because of your experience, you have probably seen all the common patterns in building a web app. As a result, you probably have a solution, or know exactly where to look for a solution to any requirement you come across. However, less experienced team would need a lot more guidance and proven best practices to follow. An analogy is a driver in a foreign country. A GPS map might not provide her with the shortest or most enjoyable route, it gets the job done. Django is great for inexperienced team. It's also great for solo consultant.

I am not sure I'd agree that Netflix could've been more profitable if they went with their own DC. I think Netflix does not utilize that much hardware. Their streaming is on their own CDN, which I heard that are placed directly in the big cable companies' DC. Tracking and storing user profiles and other meta data would not require a huge amount of hardware. I think Netflix also does quite a bit of video encoding and analytics, but these workloads would fit well with EC2.

Yes, I guess the 150 machine is more like a guideline. It depends very much on your workload. If you need 150 beefy machine running at 90%CPU all the time, then yes, running your colo may come out ahead. On the other hand, if you run a system with an online fleet of say 20 web servers, a couple of beefy databases, and some offline analytic workload, then AWS will definitely come out ahead.

And that is not considering other AWS offering. I've come to really like DynamoDB. It takes some time to get used to, but I've found it's solving more and more problems for me, without having to scale OPS engineers. There is a danger of lock-in, but I guess as I am a paying customer, it's not going away anytime soon, and Amazon is probably not jacking up the price either.

FB search is terrible because there is no incentive for an engineering team to work on it, and in fact maybe quietly discouraged from doing so. FB make money on from the news feed, so any feature that distract users from scrolling down their feed will come up losing on an A/B test where ultimate metric is ad views and clicks.

Neat! It's been a while since I wrote Java code. Java8 lambda syntax looks quite reasonable. I'd recommend the author to make "Getting Started" section a bit more detailed. One of the gripes I've heard is that Java setup and build are non trivial for inexperienced engineers. So maybe some script/screenshot HOWTOs or adding a Dockerfile?

Square root of relevance signals to "smooth" them out? I worked a bit on tuning search results, not at Google scale though and it's long time ago, but taking square root, or even logarithm of out of scale signals is a very common trick. Heck, even with the classic tf-idf signal, there are a bunch of heuristics like capping off tf, taking log of idf, penalizing doc length, etc. It's really a trial and error thing.