[marcell]

For Triplebyte they have a two hour interview with a real engineer who has passed their own interview. Even if you cheat on the live interview, you still have to do an on-site at the company that might hire you, so it’s not a huge benefit to cheat.

[nikanj]

Waste time doing an online test, for an opportunity to waste time with a faux interview, for the opportunity to..have an interview at $bigcorp.

What’s the TripleByte value add here, why should I spend time with them? Why not send my CV to $bigcorp directly?

[nrb]

1. If you are out of town and trying to relocate to SF, they handle all the scheduling, travel, lodging, and transportation for you.

2. Skips past the resume filter, HR phone screen, and technical phone screen. After you've gone through the TripleByte technical interview, you skip directly to on-site final interviews at the companies. If your resume is non-traditional (self taught, bootcamp, etc) this is pretty significant.

[throwaway0255]

You don’t get skipped past phone screens, you still have to do those.

I don’t think you skip resume filter either. They claim they don’t share the resume, but on multiple occasions I had someone from the company magically view my LinkedIn soon after I was paired with them.

It’s not difficult for them to find you, there’s not really much anonymity built into TripleByte after you’re accepted.

You do get to skip remote technical screens, which is the primary benefit imo. That might be the only benefit to the job seeker, actually.

[nrb]

Skipping the resume filter meaning making it to starting the phone screen process in the first place.

[e_b]

Triplebyte is just a different filter.

[burner87]

Using a burner because I don't want to tie this to my company.

You can send your CV to $bigcorp directly, but $bigcorp is inundated with CVs, and many CVs have a casual relationship with reality.

I do a lot of interviews on the hiring side. I look at a lot of resumes. And other than "worked in a similar position at a big-5 tech company or well regarded unicorn," nothing on a resume provides much signal. There are a handful of universities that make me pay attention, and a few particular programs outside that handful, but even that is weak signal. A candidate saying "expert in [whatever]" is useless.

I spend about 10% of my work week doing technical phone screens. This comes after a recruiter reads a candidate's resume, talks to them on the phone about their experience, and decides they're a plausible candidate. I get on the phone with them, we fire up coderpad, and I ask them to start coding. Nothing requiring exotic data structures or algorithms, just a straightforward "make a class with a couple of properties" type thing. Something along the lines of:

Part 1: Make a class that represents a playing card. It should have a suit (clubs, diamonds, hearts, or spades), and a rank (two through ten, jack, queen, king, or ace). Make a way to print out the card as a string. And make a comparison function that can tell whether two cards are the same (same suit and same rank).

Part 2: Make a deck of all 52 possible cards.

Part 3: Make a 5-card hand by picking 5 random cards from the deck.

and one or two more similarly straightforward parts after that.

There are a few places where a candidate can show off (e.g. override the default string representation for the class, anticipate that we might want a comparator that does more than just check for identical cards, or use a special data structure, use an optimized algorithm for picking 5 random cards) but none of that is necessary to pass the interview. More than half the candidates I talk to can't get through part 1. Half of the remainder can't get through part 3. Most people who get through part 3 move on to an onsite interview.

70 or 80 percent of candidates who say on their CV that they can code, and have convinced a recruiter that they can code, can't write a class and a couple of functions. So a lot of companies do the easy thing, and just toss out (or at least don't fast track) all the resumes that don't include particular schools or companies, or come from referrals. And in the process, companies lose out on tons of qualified candidates, and those candidates lose out on jobs.

If Triplebyte can reliably identify candidates who can pass a phone screen, when I'm sorting through resumes I'd be happy to treat a Triplebyte stamp of approval the same way I'd treat a degree from a top school.

[vonmoltke]

> 70 or 80 percent of candidates who say on their CV that they can code, and have convinced a recruiter that they can code, can't write a class and a couple of functions.

I work for a large non-Big-5 and our failure rate at that level is far lower. Is this only or predominantly for juniors?

[nikanj]

It's predominantly for household names. People know that scoring a job at one of these companies would be a golden ticket.

It's almost like playing the lottery. You know you probably won't win, but it's nice to have a dream.

[ajeet_dhaliwal]

70 or 80 percent of candidates who say on their CV that they can code, and have convinced a recruiter that they can code, can't write a class and a couple of functions.

Any other hiring managers here see this? It just seems hard to believe. Are you sure it's not the case that they can code just fine but just can't balance a red black tree in 5 minutes in a language that was never mentioned as a requirement before while a stop watch is shoved in their face while also having to jump through rings of fire to the background noise of some people sighing disapprovingly.

[paranoidrobot]

The number is honestly not surprising, particularly if it's a genuinely large and well known tech company.

When we've put out the call for Senior Developers experienced with language X (pick any: C#, Javascript, Objective-C), we get people who can't even write a basic hello-world statement being sent through by recruiters.

Years ago we built a basic coding test which is effectively "Here's a class, now modify it according to these sets of requirements" - something that anyone modestly competent in those languages can do in 20 minutes at most. We deliberately avoided writing trick questions, and tried to write it in clear basic language to avoid any issues with language barriers.

Applicants are asked to either do it at home, or if they don't have the ability to do so, they can come into our offices and sit in a meeting room with a laptop to complete it.

Keep in mind, this is for people who've already submitted a CV (directly, or through a recruiter) and report they have years of experience with these languages.

We've had people write back saying that it was too difficult, others who submit complete garbage that not only doesn't compile/run, but doesn't even have vaguely correct syntax for the language.

We had someone who took up the offer to do it in our offices, who we had to kick out after they sat there from about lunchtime until we were closing the offices at 7pm, and all they had done was copy down our model class from the questionnaire (incorrectly) and write some comments about how they might implement it.

I thought that it kind of overstating of skill was limited to just development roles, but having seen the quality of the people applying for Senior-level roles as DBAs, Sysadmins, BI folks - it's all terrible.

[zwily]

Years ago, while hiring people claiming to know C, at least 75% could not do a simple “reverse a string (char*)”.

[e_b]

There's also not a huge benefit to doing the test at all.