Ask Me Anything

Could it be that I am not trying hard enough to find projects that both allow technical growth and are well-aligned to team priorities? Any tips for doing that?

Companies like Meta, Microsoft, and Google are so large that there should often be teams with opportunities that better fit where and how you want to grow. This becomes harder during certain times, like around layoffs, since people become risk-averse such that:

1. People don't leave teams as much as they usually do, and

2. Managers sometimes pause hiring until the future funding picture clarifies

Despite those dynamics, the strongest people always have choices. Managers will often bend over backwards to get someone who's consistently outperformed in the past. So, to you question: a) I'm pretty sure there are other projects within Meta that suit your growth ambitions, and b) those opportunities will be easier for you to secure the better your performance track record (which is incidentally why you should do a great job no matter where you are, even if you know you're leaving soon).

In terms of thinking about what to learn / how to grow, do you see certain technical disciplines as having more 'AI longevity' than others, given the rate of how quickly coding models are improving?

People answer this question way too confidently, IMO. For instance, you might recall all the breathless articles two years ago about how the future was all going to be "prompt engineering." Or the current talk around "taste" being everything, and PMs inheriting the earth.

These perspectives have all captured elements of what's true -- but if the dot com experience taught us anything, it's that it's very hard to call the winners during times of rapid change. VCs are in the full-time business of trying to call winners, and yet the average VC performs terribly. How are people who only occasionally think about these dynamics supposed to do any better?

All that said, my intuition is that hard things will always build more valuable skills, simply because they're hard. I have a tough time imagining a lay vibecoder stumbling out of a mediocre business school doing better, on average, than someone who invests deeply in understanding how to build large scale systems with the help of AI. Hard things are often profitable if not simply because most people aren't willing to work that hard. So during times of uncertainty, I'd bet on the person who:

• Embraces instead of resists change

• Works harder than others to stay on top

• Switches their approach based on high alertness about trends of change

How have you navigated a growth/risk-reward tradeoff in your career? When did it pay off in direct or surprising ways, when did it not?

My personality is very open to new experience, so I've rarely been risk-averse. I'm much more open to change than the average person, so perhaps I've had an edge this whole time. But FWIW, I've found every major change in my career to be a positive shift. Some might say it's just my perspective (i.e. perhaps I'm making lemonade out of lemons), but I think there's a very real way in which changes, made for the right reasons, are of course likely to yield benefits given selection bias (i.e. that's presumably why you chose a specific change in the first place).

Far more importantly than choose the right change, though, is probably the wisdom to differentiate which changes are reversible vs. not (i.e. "one-way doors"). It's far more important to be sure about the latter, whereas in the former case, it behooves you to simply take action quickly and be ready to change again when things don't go the right direction.

Most of my career could not have been predicted from the outset. So in a very real way, nearly all of it has "paid off in direct or surprising ways." I truly wish you the same.

Deciding when to step away

In some ways the decision to step away from OpenAI a year and a half ago was very easy: my mom had advanced liver cancer, and I moved my parents into my house so I could take care of her full time. That was a great decision, and very easy to make because I knew it was the right choice. Zero regrets on that.

The thing I struggle with more is how and whether to get back into industry. I still love building things and am working full time on a variety of side projects. But it's not the same as having coworkers. I'm a bit unsure if joining a traditionally structured company is the best thing for me right now, since AI enables us to work in much smaller teams. Part of me wonders whether the best work environment for me might be a team of 3-5 people.

A little part of me wonders whether I should also try a completely different profession, the way I've been teaching at the University of Washington part-time for the past 8 years.

Managing intensity

I think I did an ok job at this during the period of my life when I worked a lot longer hours. But some things made it easier:

• In the period of work where I had a sleeping bag in the office, I was married without kids. So I had far less responsibilities.

• I was a lot younger in many of those crunch periods, which made daily energy less of a problem.

• I think I was objectively less effective on a per-hour basis during those sprints where I worked very long hours in the office. More effective in totality, but less effective per unit time.

• I'm not convinced all that intensity was positive. I suspect I could have reached similar career goals without pushing quite as intensely had I chosen better how to spend my time (e.g. oftentimes during my early 20's, my extra hours were spent doing things that weren't that important).

Internalizing feedback

You're right that not all feedback is good or wise. The difficulty is in differentiating the two. I think that if you only ever implement feedback you already agree with, you arguably might not need the feedback at all. We should at least occasionally try to implement some feedback we didn't at first agree with, just to witness the empiric results. But ultimately, not all feedback is good, and some feedback might be downright harmful.

I don't have any wisdom around this other than to say it helps to triangulate amongst a few people. For instance, I've recently gotten feedback from two different sources, one professional and one social, that I jump too quickly into the next topic of conversation without letting emotionally heavy things have some quiet space to sit. The fact that this came unsolicited from two different sources is a strong indicator that the feedback is consistent and accurate. I'm going to try to change this.

You sound emotionally weary, so there's an argument that the best thing for you is to leave Meta — not necessarily because it's the best path for your career, but it's sounding like the best path to helping you feel reinvigorated about your career. There's a lot to be said for feeling better about something, regardless of whether that thing is the optimal for you career-wise.

One question is whether getting to IC6 is a prerequisite for your happiness. Realize that companies like Microsoft and Meta view IC5 as a reasonable "career-long level" — meaning even lifelong IC5's, if they're solid, are well-appreciated on most teams. If getting to IC6 is necessary for you to be happy, then a hard, honest assessment by someone with full context about your realistic ability to get there will help. The rest of this answer is premised on the assumption you have the skills and abilities to reach IC6.

Here are steel-man arguments that would make each of your options hypothetically the best one:

• Keep pursuing IC6 on the same team. You have a great manager who supports you and has high confidence you'll reach IC6. You are respected on the team and your context makes you much more effective than newer IC5s. The team works in the area you're most excited about in the whole company. What the team needs most are exactly the skills you have.

• Change teams within Meta. You like the mission and benefits of Meta, but things in the previous paragraph aren't true. You have learned how to operate differentially well within the culture of Meta relative to how you'd operate elsewhere.

• Move out of Meta. You have an offer with a company which you are more aligned with mission-wise or culturally. Your skillset is a great fit with what your new team needs. The size of the new company is well-matched with the size of company you do best at. You're so discouraged at Meta that any change would feel welcome regardless of outcome.

At the highest level of abstraction:

• Don't make your happiness contingent on something you can't control. This, for instance, is why I recommend an objective outside assessment as to whether you have what it takes to get to IC6.

• There are many less-competitive companies which would happily give you a Staff Engineer title today. In fact, the weaker the company, the more they'll be willing to throw out all sorts of titles and levels. Is it really the IC6 itself that matters? Or is it something else you're trying to get?

• Sometimes we change because we know exactly what's wrong. Other times we change because we can't figure out what's wrong, but suspect that continuing isn't going to improve things. Both are fine reasons to, especially if you're not risk-averse.

My intuition is that you submitted this question because you're already emotionally ready to move on from Meta, and that you'd be happier doing so regardless of outcome. But the ultimate choice will come from you deeply introspecting to figure out whether that's indeed right for you.

A series of very hard (but fascinating) questions, in a domain I love to talk about. But first, the caveats:

• I'm (obviously) not a financial advisor.

• Even if I was, the error bars on any recommendation here will be huge. If you take the advice, you need to be emotionally resilient to massive swings that are possible, any which way, and not roil your emotions with hindsight bias.

trying to buy a home with cash because I never know when there might be a RIF

There will definitely be a RIF at some point. Whether you're included is a different question; but yes, you should prepare for uncertainty. Especially given how quickly the industry is changing.

That said, it's unclear to me whether you should buy with cash. It's smart not to be "cash-poor" after moving into a house with a huge mortgage, so it's good you're saving up, but there are other tradeoffs (e.g. the nicety of being able to customize your living space when you own vs. rent, etc). Buying in cash is definitely the most fiscally conservative approach though; then again, Warren Buffett said his biggest mistake was buying his home instead of renting for life. Some people would argue that buying, at all, is a mistake (relative to investing your money wisely), given the average US home value growth over the past 70+ years is something like 4% a year. In the end, I think you shouldn't think of a home as an investment; you should either buy one because you enjoy owning one (and view it as an expense), or rent for life because you don't want to put the majority of your assets into one real estate gamble.

What do you recommend investing in and should I automate it like my retirement? So far I have been slowly buying and holding Vanguard ETFs

I'm fiscally conservative and don't chase returns... so I'm completely behind index ETFs and manual rebalancing periodically. When you say "automate it," I'm not positive what you mean, but if you mean "follow a principled formula and don't change the approach based on emotion," I'm a big fan.

when there seems to be a dip

This sounds a little bit like trying to time the market... which doesn't work. It also sounds like the opposite of what you might have meant by "automate it," so you can guess which side of this argument I'd fall on. 🤣 I'm all for dollar-cost-averaging yourself across a principled asset allocation, and letting time-in-market do all the work.

Do you think the US stock market will crash?

Yes, always does, because it's cyclical. Might be pretty bad this time because of the US debt problem, which feels unresolvable via normal politics until something disastrous happens.

Should I focus on ETFs invested in international companies?

Unclear. I have something like 15% in international ETFs, just as a matter of principle. But it's not clear to me that you should increase your international exposure simply because you anticipate a US crash, because sometimes when the US crashes, the rest of the world does even worse. There's also a funny way in which buying US large-cap is a good deal of international exposure anyway, because so many of those companies make a huge slice of their revenue internationally. My vote is just to be principled about the amount you'll want to put into international ETFs and to stick to that, resisting the temptation to adjust the portfolio based on your prognosis of world situations (i.e. get out of the business of being an oracle).

Your approach is exactly right. As someone just starting in industry, how are you supposed to gain experience if you never look at code? It's fine and good for experienced engineers to not look at code output because they've already developed a sense for what things go wrong/etc, but how are you supposed to learn those things without ever seeing the code?

You're also right that this will make you slower in the short term than someone who ignores the code from day one straight out of college. But in several years, you will be the one who's gained deeper experience and can go farther.

The answer becomes much clearer if you look at how many non-coders get the vibecoding bug and than give up once the project is 85% done -- because inevitably, without a coding background, you'll often hit a wall at some point and be fundamentally unable to solve the issue because you lack some basic understanding. You are building yourself into the opposite of that person.