Building an engineering team, with TripleByte founders Ammon Bartram and Harj Taggar
There’s nothing more important than building a great team as you work on your company, and Triplebyte is a service that helps finding good candidates. The idea came up because the founder, Harj, was previously a partner and Y Combinator and realized that the number one problem for most startups was hiring great engineers.
- (2:00) Hiring sucks:
# It takes a lot of time even to convince someone god to talk to you, and time is a scarce resource especially for founders.
# It’s very repetitive: lots of phone calls, meetings, offers…and most will fail anyway.
# It feels bad when you like someone and they reject you by not wanting to come to work for your company.
(4:11) Where to source engineers from, from the best to the worst option:
- (4:40) Personal networks: when you’re small, hiring is not just about technical skills…you must find people that work well with you, which is why hiring someone you know already (or someone that worked with people you trust) is the best thing and reduces the risk. People often don’t ask friends because they’re afraid of being rejected. They are also afraid of offering a job to a friend knowing that the startup can fail. But still, do make a list of all good engineers you know, and ask. Keep pushing until you’ve at least shown something to them. If they still say no, at least ask them to recommend someone else, and start over with the process. Then tap into your employees’ network, for example by asking them to put names on a spreadsheet, and offering a referral bonus.
- (9:08) Hiring marketplaces: there’s more demand then offer, and so candidates get a lot of inquiries and decide who they want to speak to, so it’s very competitive. The fee is usually between 15–20%, so quite expensive. Because most candidates are actively looking for a job, this is a good option to hire quickly.
- (11:55) Linkedin (and Github): used by technical recruiters that send out hundreds of messages a day, but the conversion rate is very low. For a startup is better to send fewer but more targeted and short messages that show that you’ve read their profile. Sign up to the recruiter plan on LinkedIn to be able to pull out candidate’s email, and use that instead of internal messages, because emails have a higher response rate.
○ (13:56) Job boards: StackOverflow Jobs and AngelList. There’s more quantity than quality, most applicants will be under-qualified. Reading all CVs you’ll get and finding the few good candidates will require a lot of time. Make the job description appealing and more personal, avoid corporate language. If there’s anything unique about your culture or tech challenges, write it, it will help you stand out.
○ (15:49) Physical meetups: not very effective, most meetups don’t have a lot of people attending, and many are there for the free food anyway. Focus on technical meetups and not corporate. You can also host one at your office, but mostly to build a network than for hiring.
- (17:27) Hire your first engineer by yourself before having a recruiter, and when hiring is starting to take more than 50% of your time. It will be like pitching your startup to convince people to come onboard, and practicing that skill is always useful to understand what resonates better about your company.
- (19:38) Different types of recruiters: contractors (you pay them by the hour), in-house, and agencies (this last type has a fee of 25–30% of the hired person’s annual salary). Start with a contractor that just sources candidates and organizes calls for you, but then do the calls and the pitching yourself. When that becomes too much work for us, upgrade to a full time in-house recruiter and train them to also do the pitch calls.
- (23:16) A usual hiring process is: resume screening, recruiter call for culture fit, tech phone screen, optional take-home project, 3–6 on-site interviews in a single day with engineers, and final decision meeting between everyone involved in the process. Companies make an offer to 2–8% of people that apply. It’s interesting to note that no candidate passes all interviews, most pass some and a few pass most. (47:10) For small companies (Series A and smaller), be less aggressive at screening people out early, do more on-site interviews.
(26:14) A challenge in designing the interview process is consistency, having interviews that are repeatable. They found out that there’s a lot of disagreements between different interviewers of the same candidate. A trial period would be more effective, but top engineers are not willing to do it. How to reduce noise then?
- (29:45) Decide what skills matter (productive or slow but careful? strong in computer science algorithms or knowledgeable about Linux kernel? are more about hard tech problems, or building a product?). If you don’t decide this, each interviewer will decide for themselves and they will reject people that you might have wanted instead. (47:20) If you’re not sure what are the best skills, focus on productivity and ownership, at the expense of code quality.
- (32:10) Structured interviews (constant process with the same questions, with a specific skill to evaluate, rather than a global feeling) are better are predicting whether the candidate is good for the company, even though they feel less good that free-form interviews. It’s better if one person or group collects the notes from all interviewers, and makes the final decision.
- (35:25) Ask better questions. Avoid questions that require knowing a very specific algorithm and prefer ones that can be solved with a series of more straightforward steps. With multi-step problems you can help with one if the candidate gets stuck, and then let them proceed with the rest.
- (38:15) Give the candidate a maximum of 3x the amount of time you think it would take you to solve the problem. The question might be harder than you think, especially during an interview.
- (39:08) Ask four or more questions.
- (39:30) Give candidates a problem, give also the solution, and let the candidate implement it.
- (39:56) Ignore credentials (school, previous companies) during the interview. They are meaningful, but don’t biased because of them. So hide them from interviewers but consider them for the final decision.
- (41:28) Bad hires are bad and costly, and it damages the morale if you then fire them.
- (43:15) Look at their best skills, not weaknesses. Everyone will look stupid on some questions, so don’t reject a candidate just because of that.
- (44:05) Ensure that every candidate likes your company even from the interview process. Reduce their stress by letting them use their own laptop and use their language, and of course be nice. (54:35) Make sure the interviewers is familiar with the tech questions they ask, and that they don’t try to sound smarter than the candidate, because these are the two things candidates complain the most about.
- (53:40) Speed in hiring is a huge advantage over bigger companies.
- (55:04) Be prepared to talk in detail with the candidate about the culture of your company. Take some risks and also talk about the trade offs of your culture; this might turn off some candidates but it also differentiates yourself and it might help you close the offer with candidates that like you.
- (56:52) Make clear offers with all the details: base salary, signing bonus, health insurance plan, vacation policy. In case of equity indicate the total number of options and shares, the exercise price, the valuation at the last round of funding.
- (59:19) To compete with giants like Google or Facebook by emphasizing learning and mentorship (at smaller companies you have more chances to experiment and make mistakes) and career progression (if you join a startup at a early age, you can also end up in top management at an early age). Also stress that the opportunity to join your company is unique and won’t happen again, while the tech giants will still be there in the future if the work at your startup doesn’t work out well.