My Journey from Econ PhD to Tech — Part 5: Q&A + Miscellaneous

On Dec 15, 2020 I posted this series on Twitter and got a lot of replies — on the one hand I was glad that it was helpful, on the other hand I was surprised to see these sorts of info weren’t more public before. Based on the many questions I got, I decided to add a running Q&A part to the series, to answer any question people might have.

Links to other parts of the series:

Part 1: Interview prep + Networking

Part 2: Interview experience

Part 3: Wage negotiations part 1

Part 4: Wage negotiations part 2 + Decision

Q: Is it legal to talk about compensation openly?

A: Yes it is! After the series was published, more than one people reached out to express their concerns. In fact, there’s Federal Law protecting employees’ right to openly talk about their compensations. And if a firm has a policy against employees talking about such issues, the firm’s policy is by default invalid.

Q: Isn’t there a Stanford policy against deadlines before 11/15?

A: You should absolutely use your university’s resources! Many universities have policy where if a firm uses a university’s event to recruit its students, they cannot enact offer deadlines before a certain date. But if you applied to the job outside of the university’s recruiting event, things might not apply — you can check with your university’s career services on the details. For Stanford it’s BEAM.

Q: What’s the recruiting process like in tech?

A: In tech, the recruiting process is typically:

• Information call with either the hiring manager (if you went through a referral and your CV was identified by a hiring manager) or recruiter #1(if you went through the generic pipeline and passed the CV screening round) — what does the firm/team do, what the interview might ask for, etc.
• Data challenge
• Recruiter #1 (the one in charge of everything until phone screen) and a coordinator (someone who’s in charge of scheduling) help to schedule your phone screen #1 and phone screen #2 (before the phone screen they should send you prep material — if not, you should ask)
• Recruiter #2 (the one in charge of onsite and everything onwards) and a coordinator help to schedule your onsite (before the phone screen they should give you a prep call — if not, you should ask)
• Recruiter #2 calls to tell you you have a verbal offer (without numbers)
• Recruiter #2 calls to tell you the numbers
• Negotiation round 1: asking for an increase by sharing your ask with them
• Negotiation round 2: share your highest counter offer(s)
• Negotiation round 3: ask-and-sign ‘if you give me x more I’ll sign’
• Sign

Of course not all recruiting processes include all of the above — it’s a superset of things. And there can also be exceptions — I’ve heard of people having 3 rounds of phone screens.

Q: How much coding do I need to know? What’s a coding interview like?

A: First, if you only know Stata and nothing else, Amazon is the only tech firm that will embrace you, so pick up either Python or R if you can. (But seriously, if you’ll stay in tech for the rest of your career, knowing coding is valuable beyond just interviewing — there are people who started at Amazon and can’t switch jobs because they only know Stata)

Second, the amount of coding you need to know to get the job varies by firm. Firms will typically send you interview prep material before your interview, so you would know how much coding the interview involves. But of course, it’ll be too late to prepare then.

I’d say, if you know everything I listed here, you’re fine. And in fact, it’s not so much about remembering these commands/algorithm/data structures, but that once you have mastered these, you’re at a level that can solve most coding questions that come up in data science interviews.

In other words: in a data science interview, they won’t ask you to write a function to implement ‘heap’, but they will ask you a question, which, in order to solve it, you will need to use the data structure heap (there are of course other ways to solve it, but knowing how heap works and using heap in your solution makes solving that problem very easy). — I’m using ‘heap’ as an example… not saying that all coding interviews involve heap if that wasn’t clear

All data science coding questions are ‘problem solving’ questions: the interviewer tells you a problem, e.g. identify all the words, which, by changing one letter, can become a palindrome. You usually have ~45 min to solve it.

I would start by asking clarifying questions, e.g. what’s the form of the input — is it already split into a list and each item is a word or do I need to parse it myself; or, by ‘changing’ what do you mean — can I delete or add letter or can I only replace a letter by another one, etc.

Then I’ll start from the brute force solution: ‘I’ll first write a function to identify palindrome, i.e. return True/False given a word, then write another function to loop through all letters of a word and given each letter, try to change it to each of the other 25 letters, and use the first function to determine if the new one is palindrome’.

Clearly it works but it’s way too slow, so usually you won’t write that out, but instead you’ll start to think about ways to improve it. Then I would suddenly recall that, for any string-operations problem, a typical solution is to use ‘pointers’ — this is something you’ll know once you have gone through enough Leetcode problems (themes start to emerge!). I would use 2 pointers, i and j, and a macro variable to keep track of the number of ‘differences’. i starts from the first letter and j from the last, and in each iteration, if i and j are the same letter I move i one position to the right and j one position to the left; if they are not the same, I increase the macro variable by 1, and let i and j move to the next position. At the end, if the macro variable is only 1, it means there’s only 1 difference, and this word can be changed to palindrome by changing 1 letter; otherwise it can’t.

And at the end you should run a few test cases to show that your function works and/or debug by test cases — you’re almost never bug-free the first time!

There’s another completely different type of coding interview where you’re given a dataset, e.g. some user pre-existing characteristics and results of an experiment, and you’re asked to answer a few questions by plotting graphs / running regressions. That should be easy for econ phd because that’s what we do everyday (except for theorists…?) But the only problem is that — you need to do this in Python or R, instead of Stata. So get comfortable doing the same thing you used to do in Stata with Python pandas or R’s equivalent package.

Q: Is referral necessary? Can’t I just apply through the AEA?

A: I can’t emphasize this more but referrals are really essential if you’re serious about getting a (good) job in tech. Here are some stories:

A friend of mine on the market right now applied through the AEA and got desk rejected from all tech firms and messaged me about it. I told her don’t worry I can refer you to a few firms and you can get un-rejected. I wrote a warm email to introduce her to one of my contacts at one of the firms she was rejected from. A few days later she received the invitation to interview.

More than one (non-econ) friend of mine applied to firm X and didn’t get a response. Given my connection with them through my interview process, I just wrote an email to the recruiter cc’ing them to say they’ve applied, and usually within the next few days they get the data challenge.

And of course, my own stories where I got rejected from Robinhood and then un-rejected because of a friend.

So why do you have to use referrals if you’re serious about tech?

• Tech is in too high demand nowadays. Among intern return offers + warm/cold referrals they can already fill the whole cohort, so why would they go through the thousands of apps from AEA?
• Signaling: for the firms that do go to AEA, they can’t tell if you’re serious or not, and given how popular they are, why would they waste time on someone who sees tech as a back up? They would go for people who have tech internship(s) on their CV, or those who went through referrals — referrals are a signal that you really spent the effort to figure out how tech job applications work and paid the fixed cost of networking
• Vast majority of tech firms don’t go to AEA — how are you going to get those if you only apply through AEA? For example, many of my econ phd friend either interned or worked at Google (data science, not econ), Quora, Coursera, Robinhood, Stripe, Zillow, to name just a few, but these don’t always go to the AEA (unlike Amazon and Uber). Not to mention firms that are even more tech-y, e.g. Doordash, Yelp, Spotify, Airbnb, Netflix — they are all major tech firms, but you won’t even be able to get in contact with them if you don’t use referrals.
• Most econ phd’s CV look too non tech-y, so if you just throw your CV at the above firms’ career website, it’s hard for them to pick it up. One way is to change your CV, e.g. add 3 ML projects in a month, but if you don’t have to do that, referrals work too (of course the best is if you both fix your CV and use a referral)

Q: How do I find referrals? I’m a lonely grad student cut off from the real world since a long time ago

A: In the real world, people are a lot more welcoming/friendly/warm-hearted/open-minded than in academia! People send and receive cold email/messages all the time, and people are used to asking for and giving referrals — they get a cash bonus if the candidate they referred end up getting and accepting the offer! So don’t be shy — you’re doing them a favor as much as they are for you :)

• Use the older students in your program/university who went into tech / Ask your advisor to introduce you to their students who went into tech
• Use your friends outside of econ who went into tech — I usually just blast on my social media to ask any question and always get a lot of replies — people are more willing to help than you think!
• LinkedIn cold message people. How to find them? Search for the name of a company in the search bar -> click into that firm’s profile on LinkedIn -> go to ‘people’ -> now you can further search by keywords like ‘machine learning’
• Use family connections (why not?)

Q: What share of your applications went through referrals?

A: I only applied for firms where I have a direct referral except for Instacart.

[firm]: [position], [outcome]:

• Google: Data Scientist, PhD university graduate, offer
• Facebook: Research Scientist, Economics, Novi, onsite
• Amazon: Economist, SCOT, onsite
• Uber: Data Scientist II, offer
• Zillow: Applied Scientist, Machine Learning, offer
• Quora: Data Scientist, offer
• Coursera: Data Scientist Semester Intern, offer + ‘don’t know 2021 full-time headcounts at the moment’
• Robinhood: Data Scientist, online assessment
• Bytedance: Data Scientist, CV stage
• Wayfair: Economist, ‘don’t know 2021 headcounts for new grad yet; follow us in AEA’
• StichFix: Data Scientist, ‘no headcount for fresh grad at the moment’
• Stripe: Data Scientist, ‘check our career website for fresh grad openings’
• Yelp: Data Scientist, ‘It was great talking to you!’
• Instacart (the only firm I applied without a referral): Economist, ‘timeline doesn’t work out — we want someone who can start now’
• LinkedIn: ‘firm-wide hiring freeze at the moment’
• Roblox: Data Scientist, ‘we are looking for people with experience’
• Doordash: saddest story in my recruiting journey — someone replied Anthony’s Twitter; I sent my CV to him asking for a chat and didn’t get a reply; a few days later I received a rejection email for their ‘Senior Data Scientist’ position…(if only he replied and we talked, he would have known that the relevant position is fresh grad not senior data scientist)

Apparently referrals don’t always work:

• If there is firm-wide hiring freeze
• If the firm doesn’t have full-time headcounts at the moment
• If the firm wants to hire someone with work experience (and you are a fresh grad)

…then referrals won’t get you in.

But if there are openings for your profile (i.e. (econ) phd fresh grad), referrals can un-reject you and/or speed up your process

Also don’t take non-replies / rejections personally — Would you expect all the guys/girls in the world to propose to you? If not, why would you expect all the firms/positions in the world to be interested in you? Job hunting in industry is as much a fit problem as marriage — if you don’t get offended when not all guys/girls are interested in you, why would you do so when some firms/positions are not interested in you?

Academic jobs are on the whole vertically differentiated — it’s hard to imagine someone who has Harvard to take a community college’s offer instead. But in tech, things are mostly horizontally differentiated, and it’s reflected in people’s choices: Haseeb took Airbnb over Google; my friend who referred me to Roblox took Roblox over Google and Amazon; a friend of mine took Zillow over Microsoft, etc. — Different firms do different things, e.g. if you’re just not interested in social media, you might not want to go to Fb or ByteDance; if you’re just not interested in 2-sided markets, maybe avoid Uber.

Q: If I’m not from a top econ program, or not a phd, what bargaining powers do I have?

A: I’d say my bargaining powers came more from my unique fit with the team than from my background per se, and it’s really on you to extract that piece of info.

In the recruiting process, it’s common to have an information chat with the hiring manager before the interview process, and to have a lot more chats with various people on the team after you got the offer. I would ask ‘what projects are the team working on / what are the big next steps or bottlenecks / what projects will you assign me to work on’. From their answer, you would be able to tell how unique a fit you and the team are:

at Uber they said they’re trying to consolidate different products created by different teams that potentially serve the same purpose, and I realized I’m a great fit because I’m a macro person — I can write down equilibrium models to study such questions; at Zillow, they used to be using purely ML to solve the home pricing problems, and now they’re trying to hire someone with domain knowledge in housing, and I realized I’m a particularly good fit because I studied housing in my grad school.

I also asked the recruiters for my interview feedback — some will share it, some won’t. One of the recruiters who did share it said I got very little negative — maybe one of the only candidates with such little negative comments based on what she has seen so far — this made me realize I rank very high and have a lot of bargaining power.

Miscellaneous

Figure out your preferences before walking into the negotiation room

One of the most important takeaway I got from my negotiation process is that, you need to form a clear and strong value system before walking into the negotiation:

• What matters to you? Money, success, fame, status, family, freedom, fun… these are all valid answers, and only you can decide
• What’s your career goal for the next 1–3 years? 3–5 years? 5–10 years? for life?
• What’s your life goal? e.g. getting married, having kids, buying homes… and how would that interact with your career decisions?

These are things no one can tell you, and there are no right or wrong answers for, and only you can decide for yourself.

Over the course of my grad school, I noticed that a lot of the frustration or discontent comes from the fact that I didn’t have a clear mind on what I wanted, and was ‘told’ to believe something I later felt uncomfortable with.

I went into grad school because I enjoy understanding how the world works. I came here to study the questions I’m curious about. I came here for knowledge and truth. But unavoidably, just like any other field, there’s a prestige ladder, and the longer you are in the field, the more likely you’ve been brainwashed to believe that the ladder is what you should pursue. All of a sudden, instead of thinking about what questions will make me happy, I was thinking about what will ‘place me well’, and I felt nervous and discouraged on the thought of a bad placement.

Then I realized how wrong these feelings were — I was pushed into believing something I didn’t set out to want to believe in. Much of the discomfort and frustration is because I didn’t hold on to my value system.

That’s why I did my soul searching before I made the decision which firms to go — I realized that what excites me is working on a start-up, because that’s a perfect combination of both (1) I get to decide what I work on and (2) It’s a form of organization that can be used to solve social problems — both of which are what I value the most in my life.