papercruncher

San Francisco | Senior Data Engineer | Tubi TV | Full-Time | On-Site | VISA (transfer H1/O1)

We are making premium TV shows and movies available for streaming everywhere and to everyone, 100% free.

Join Tubi TV and reinvent the way consumers discover and consume premium content. With over 40,000 movies & tv shows, Tubi TV has the world's largest catalog of premium content, all made available to consumers for free. We offer very competitive pay, full medical, dental & vision benefits, catered lunch, gym subsidies and your choice of hardware

We are hiring for:

- Devices frontend engineer: Hack on the web, xbox, roku, TVs and anything that can stream video. If you have a good eye for design, love working on the latest web technologies or squeezing the last bit of performance out of a hardware constraint Smart TV, then this position is for you. More details: https://tubitv.recruiterbox.com/jobs/fk06f7q/?referer=hn

- Senior Data Engineer: Come and build out the next iteration of our data platform. Full autonomy and end to end ownership. Work on your choice of algorithms using Spark/Flink or any other stacks you deem suitable. More details: https://tubitv.recruiterbox.com/jobs/fk06xpn/?referer=hn

If you'd like to chat more first, shoot me an email and mention HN.

Lead iOS Engineer | ONSITE | Full-time | San Francisco, CA | VISA (will transfer H1/O1/E3)

We are looking for a Lead iOS engineer to take Tubi TV to the next level on iOS and tvOS for Apple TV (and Android and xbox and a bunch of other devices if one is also interested in that). This is a hands on position, but you will get the chance to shape & grow the team and install best engineering practices. You would also of course be deeply involved in the product planning and decision making. Tubi TV is the largest free streaming TV network; think Free Netflix. We are VC funded but we also have substantial revenue and inked partnerships with major Hollywood studios. If you enjoy consumer products, movies, good UI/UX (we have two designers in the team) and a professional work environment without ninjas or rockstars, you will have a lot of fun at Tubi Tv.

Interview process is: 1. Intro call to learn more about the company & the role (optional) 2. Technical screen over Google Hangout (any language/editor you want) 3. Onsite interview, ~1/2-day

To apply: https://tubitv.recruiterbox.com/jobs/fk0hkpt?referer=hn

Lead iOS Engineer | ONSITE | Full-time | San Francisco, CA | VISA (will transfer H1/O1/E3)

We are looking for a Lead iOS engineer to take Tubi TV to the next level on iOS and tvOS for Apple TV (and Android and xbox and a bunch of other devices if one is also interested in that). This is a hands on position, but you will get the chance to shape & grow the team and install best engineering practices. You would also of course be deeply involved in the product planning and decision making.

Tubi TV is the largest free streaming TV network; think Free Netflix. We are VC funded but we also have substantial revenue and inked partnerships with major Hollywood studios.

If you enjoy consumer products, movies, good UI/UX (we have two designers in the team) and a professional work environment without ninjas or rockstars, you will have a lot of fun at Tubi Tv.

To apply: https://tubitv.recruiterbox.com/jobs/fk0hkpt?referer=hn

TubiTV | San Francisco | ONSITE | VISA | Senior Front-End, Backend, iOS and Android Engineers | Full Time

Love TV & Movies? Tubi TV is hiring across the board. We recently signed content deals with major Hollywood studios such as MGM, Paramount Pictures, Starz, Lionsgate, ITV, and Hasbro Studios. We also just closed a Series B round of funding and have healthy and growing revenue. Roles we are hiring:

- Senior Backend Engineer (NodeJS) - Lead Android Engineer - Lead iOS Engineer (Obj-C, tvOS) - Senior Frontend Engineer (html5/css/vanilla javascript, React/Redux optional)

Send me an email (marios at tubitv.com) with your resume, or any questions you may have.

I was one of those affected by the mentioned data loss and inconsistency issues. Even though I kept backups, I was fairly annoyed at the time, but now I'm glad to see they moved past it and things are stable.

Does Chef Watson count? It's actually pretty neat as a consumer tool, it suggests the weirdest things but somehow they taste fine. No idea about the inner workings though

https://www.ibmchefwatson.com/

One more upvote for Peter, a great guy and a fantastic attorney

Always exciting to see my home country (Cyprus) mentioned although I'm 90% sure that's just a shell corp and the address is their accountant's/lawyer's office.

Glad to see a list that doesn't just rehash the same books/papers. The Purely Functional Data Structures suggestion looks like a great read

I second the recommendation, it was fairly short, an entertaining read, solid tips and no fluff

Sort of. You can definitely be up and running relatively easily, it's just there are so many different knobs and switches and ways to set things up, you never know if you are doing things the right way. I also agree that the documentation is not particularly helpful and sometimes it's contradictory. The software itself is rock solid though, gives me very little trouble (other than logrotate & emperor mode being a pain)

Sweet! The source is very nicely laid out as well, can serve as a realistic example of building a modern ReactJS app

Yes, sadly there are. It's not necessarily out of malice, a lot of folks just don't grasp how big spider traffic is. Even Google Analytics doesn't filter out crawlers by default, that setting is buried somewhere in the admin panel.

You're right, "time spent on ad unit" is simplistic but it's not inefficient. It shouldn't be the only thing in your arsenal but it's a good signal.

I reckon you are also overestimating the amount of sophistication that goes into these systems. The ones that are out for purchase in a black market become well-understood within a few weeks and are easy to fingerprint. Since the network is paying on net-30 days, there is some time to retroactively run some analysis and filter out provably non-human traffic.

I work for an AdTech company. One of our selling points is that we track much time users actively spent on our ads (= ad is in viewport + tab is active + user has moved the mouse or scrolled in the past 20 seconds). If we consistently see a source of traffic spending < 30 secs, then we manually start digging into the logs.

If you're a brand/marketer, I recommend two easy things to increase your ROI: (1) Make sure you are not being charged for GoogleBot traffic. Seriously, for frequently updating sites, self-advertising GoogleBot can account for up to 10% of all impressions. Ad Networks can easily filter that traffic out. (2) If you're paying per impression, work with partners that charge on viewable CPM.

I took Phil's class (6.171 - Web Development Lab) in college. It was a ridiculous experience, regularly had to put 40hrs/week just on this class and I'm not exaggerating.

Story time: As part of the class we were working with real clients to develop a web project for them. One day during class, he talked to us about the "technical sidekick", typically a friend of the client's that is somewhat technical and is behind-the-scenes advising the client and typically contradicting all your architectular choices as a developer. Well, to noone's surprise, a few weeks later, our client rolled into a meeting all of a sudden knowledgable as to why .NET was inferior, SQL Server bad, etc. We later found out that he took the time outside class, to take the client out for dinner and act like an adversarial technical sidekick just to teach us that lesson. I was pissed off at the time but now deeply appreciate that lesson

Correct, the problem I was facing prior to this announcement is specific to mobile web (i.e. open download link from browser)

Hopefully we'll finally be able to accurately track download attribution on mobile web. Currently, if you need to track the performance of promotions/ad campaigns for free apps on mobile web, you need to use costly 3rd party services which use "fingerprinting" techniques of dubious accuracy

I replaced most of our custom models (build on scikit-learn) with this.

Pros:

- Super simple API, you can be up and running in a few hours

- I don't need to maintain machines/code to do all the training and querying (huge win for us)

- Relatively fast

- Very limited ways to influence the models it builds

Cons:

- Like other Google APIs, it 500's from time to time (~0.5%) for no good reason

- It's a big black box, doesn't expose much information about the models it builds other than a confusion matrix.

- Very limited ways to influence the models it builds

If ML is critical to your business and you know what you are doing, you can get better results with a custom stack.

Don't hate yourself, interviewing is a skill that can be improved with practice. I'm not hiring right now but if you're in SF, I'd be more than willing to give you a mock interview and all the honest feedback you want. Email is in my profile.

Switched to ZenPayroll from Intuit Payroll and couldn't be happier. ZenPayroll is a great product starting from the onboarding experience, to running weekly payroll, to their customer support, even the automated transactional emails they send out. Intuit on the other hand was so terrible, it actually cost us money because it buried some critical action items (local tax filing forms) under some obscure setting

> Apparently all we have to do is stand back and let the dominant players commit suicide while they dance around Schroedinger's Cash Box

Don't know how true this is with regards to MPEG, but I really enjoyed this quote

The linked page on the tweet offers great advice, the only two things I can add are:

* Take a sheet of paper and write down all definitions for easy reference

* If you are planning to use the learning from the paper in a production system, make sure you fully understand all stated (and unstated) assumptions the researchers made when presenting their results.

I have never in my life seen a VC look at source code. Even the more technical ones that might be interested in the technical details limit themselves to the overall architecture (and even that mostly out of curiosity). Is anyone else's experience different?

I know it's not the core topic of the article, but would love a deeper dive into your Kinesis implementation. What is your API & processing servers written in? How easy was it to pump events into Kinesis? What does your event processing architecture look like with respect to Kinesis? Where does the event data end up? RedShift?

[unrelated] As someone who used to rack a lot of 120mm mortar shells, I empathize

I worked in Vancouver as a Microsoft employee for a few years, it was a great experience. Ping me if you want to chat

Tempted to deploy via rsync on my laptop

We switched to the google pagespeed module for nginx and stop worrying about it. It doesn't do minification very well but it does concatenation and takes care of a bunch of other stuff as well, like image resizing, CSS concatenation etc.

We used to use the HTML5 Boilerplate script but it was horrendously slow on EC2

I hate to be the guy that just critizes, but I can't find anything that I like about the redesigns. Take Yahoo Sports for example: the dark background makes it hard to read anything, it's crazy cluttered and even after filtering for NFL I get this stream of useless news. I'm guessing that they did some research and discovered that their core audience likes cluttered designs and the whole image-first, clean design paradigm is not a good idea for content sites.

Also, why doesn't the blog announcement link to the actual sites???

That system breaks down if you start dating from within your social circle (friends of friends). I know you'll tell me it's a big no-no but you'd be surprised how much more common it becomes as you get older