I swear, I'm not building another Triplebyte

Triplebyte was a hiring platform that interviewed software engineers. Those that passed were introduced to companies and could skip the regular interview process.

In my last post, I mentioned I’m pivoting my efforts. I am building a hiring platform and interviewing engineers for it. Those candidates can skip the regular tech screen for companies I introduce them to.

Perhaps the key difference between the two is my philosophy and methodology on interviewing.

My Triplebyte experience

I was job searching for much of 2018 - 2020 and I took the Triplebyte quiz a few times. Surprisingly, I managed to pass the initial multiple choice quiz for their backend track rather than their frontend (I made a few lucky guesses). I proceeded to then do their one hour live assessment and completely bombed it.

The Feedback

Ouch, it’s pretty rough. This feedback actually sounds like something I would write for the candidates I interview.

I’m pretty surprised I did so poorly in the coding portion. Algorithms and live problem solving have always been my strengths. As is communication, but what most likely happened is that I had no idea how to build real software and that was apparent when speaking with me. But not being comfortable in JavaScript is truly shocking!

The rest of the feedback makes a lot of sense. I only had about a year of engineering experience and almost no backend experience. I really didn’t know how to build software. To this day, I could stand to learn more about databases. Even though I had taken some CS classes at this point, my understanding of low level systems was only high level. I’m still not great at quickly grokking new codebases.

The Multiple Choice Quiz

I recall in their frontend screen, they had a questions like these.

What is the order of logs this code will produce?¹

console.log('A')
setTimeout(() => console.log('B'), 2000);
setTimeout(() => console.log('C'), 0);
console.log('D')

This is a decent question to test your knowledge of JavaScript code execution. And engineers who know the answer are probably pretty good engineers; I’m confident that the best JavaScript engineers I know would get this question right.

The problem is that this is a silly snippet of code. No one would ever call setTimeout with 0 milliseconds. If they wanted code to run later, they would just move it down the file. It’s a good educational question but it doesn’t tell me anything practical about an engineer’s skill.

I’m nitpicking since this is just the Triplebyte intro quiz, but this is how they started filtering people out at the beginning steps of their process.

The Pop Quiz

When I got to the onsite, I remember a live Q&A section. Two questions I recall. Explain how malloc works (or maybe what is malloc) and what is a bloom filter?

I haven’t written enough C in my life or studied enough computer science to really recognize what he meant by malloc, though I did understand it at a high level. Bloom filters are a somewhat niche data structure. They don’t have nearly as many practical uses as the more common ones.

I doubt someone who didn’t know these things would automatically fail the assessment, but I’m a bit confused why these quiz Q&A type questions were included.

My assessment of their assessment

Looking back on it now, there is a disconnect between how they actually tested engineers and their ethos: “anyone regardless of background had a chance to pass their assessment.” I remember how they promoted stories of self taught hobbyist engineers who found jobs through them.

I do think the Triplebyte assessment is an indicator of if you were a good software engineer, but the the vast majority of people who did pass had years of professional experience.

What I’m doing differently

This retrospective on Triplebyte is a phenomenal read. The tl;dr on why Triplebyte failed is as follows:

• They tried to scale their product to appeal to VCs and it became less personal

• In the effort to scale, they neglected user privacy

• They tried to scale their product and transformed it into something that did not have product-market fit. Their customers wanted top tier candidates who came from prestigious schools and company. Triplebyte was no longer bringing in as many top tier candidates.

I’m not taking VC money

Triplebyte had to scale because it was funded by venture capital. It would have succeeded as a lifestyle business but that was not to the interest of the venture capital backing it or its founders.

Thankfully, starting an agency recruiting business is cheap. I estimate that my first year costs will be less than $50K. Most of that will be SaaS software like a code interviewing platform and document signing service. There may be some legal and accounting services too that add up³ .

Privacy is paramount

Thinking about the backlash Triplebyte experienced with their anti-privacy product pivots, privacy controls is one of the first features I built into my platform.

My product: evaluating for potential

My biggest gripe with the TripleByte exam is that it does not measure for growth. Sure it’s background blind but it is also blind to potential. You have to know a lot about building software already to pass their test.

This criticism applies to a lot of interviews I’ve done in the past too. In some ways, I think what Triplebyte did was no different than what FAANG+ companies do by filtering with a perhaps unnecessarily high bar.²

When I evaluate engineers, I’m not only looking for people who can perfectly answer the question. I’m looking for those who have the right instincts on solving the problem; who I can guide to the solution.

I talk a bit about how to write a question in such a way that both a junior and senior candidate can pass in this post below.

I’m transparent with the companies that hire through me; you won’t find many ex-FAANG engineers on my platform. Frankly, those engineers don’t need my platform. As someone who fits the description, I’m constantly getting DMs from interesting companies. If I was job searching, I would take these calls. I wouldn’t benefit much from a hiring platform.

I’m looking for engineers in the middle. The diamonds in the rough.

Back in 2018-2020, I came close to offers from a number of startups that I was really interested in. I think there was a high likelihood I would have taken those offers had they materialized.

It’s impossible to say if I would have been as good of an engineer had I worked at startups from 2020-2025 instead of working at Square. I certainly would have learned different engineering skills but I suspect I would still be the same person.

Companies should take a chance on my pre-vetted engineers. You can snatch them up before they do the algorithm grind and join a FAANG company.

I have yet to determine if my hiring platform’s ethos matches with what the market is looking for, but early testing is promising.

Want to hire through me?

I test for more practical engineering instincts and skills and propensity for growth. I still incorporate algorithm and data structures into my questions, but I put less emphasis on that type of problem solving. The engineers who have passed my interviews are good practical problem solvers and collaborators.

If you’re a company looking to hire engineers, hopefully I’ve convinced you by now to hire through me. I’m looking forward to us working together to meet your hiring goals.

1  The answer is A, D, C, B. The setTimeout callbacks always run after all other code execution in the scope is done; even if the timeout is 0 milliseconds.

2  FWIW I do think Triplebyte’s assessment tested for more practical skills than interviewing with difficult algorithm questions.

3  And if I get sign-off from my accountant, I’m totally buying one of these for my home office as a business expense. I always drink water when conducting interviews; I can’t leave my desk during these calls to refill my glass.