(*) I'm frequently asked how much time this involved. I probably put in 8h/day over that summer, then maybe 6h/day during the school year (I self-studied on the sly during school -- I typically I had to hide what I was doing, act like I was paying attention, and keep an ear out in case I got called on), then maybe 3h/day the following summer when I started leaning further into my own projects, and then the following school year I gradually transitioned to focus fully on projects and research investigations.

However, I was NOT engaging in the most efficient learning techniques (though I was spending a lot of time solving problems, which was good). So, as a rough estimate, maybe 8x90 + 6x270 + 3x90 + 0.5x3x270 ~ 3000 hours, which could probably have been compressed into ~1500h if I had been properly using the most efficient learning techniques (as properly as can be done manually by a learner themself), and maybe something like 750h if I were working on an an adaptive learning system that leveraged those efficient learning techniques to their fullest extent and took care of all the work behind the scenes?

(**) I didn't submit any sort of portfolio evidencing my self-study or research; I didn't ask my science fair mentor professors, my university course professors, or even a single math teacher for a recommendation; I only took 2 AP exams before the application cycle and waited until afterwards to take many many others; I didn't source any feedback on my essays; and I didn't make my quantitative passion / self-study / research the main focus of my essays (I wrote mainly about community service instead -- I spent several hundred hours working and later volunteering at a local program for special-needs kids). I even paid money out of my own pocket to take Linear Algebra, Multivariable Calculus, Intro to Proofs, and Organic Chemistry at community college after already having self-studied the material, just to get those courses officially on my transcript -- and besides the transcript, I didn't even mention it on my college applications.

Why not? I guess just didn't know any better. There is not a single person in my family who is mathematical, technical, scientific, or academic, and the sole contribution of my high school counselor to my college applications was misspelling the word "Calculus" as "Calculas" on my transcript -- so proper guidance was sorely lacking, and I was too engrossed in my actual learning & projects to pick my head up and show them off, write about my inner drive and passion for the material, or even notice the lack of guidance. But you know what, when you miss one opportunity, all that means is you learn from it, work harder to create another opportunity, and work smarter to capitalize on it.

(***) I'm frequently asked, with skepticism, how that was possible. How did I get the job, and how did I make it work logistically? It's actually pretty simple. I got the job because I did great work as an intern. I mean, that's typically how it goes, right? You do great work as a student intern, you get a job when you graduate. So, it shouldn't be a surprise that there is some level of "really great work" that can get you a job BEFORE you graduate. The best way I can describe that level is by saying that during one of my analytics presentations, the CEO exclaimed "More than 3 out of 4 analysts I've met in my life could not do what you just did. How old are you, 20? What, 19?! Jesus." When you perform at that level, doors open up for you that everyone else perceives to be solid concrete walls. (For the skeptics: I should also mention that I had no family or friends affiliated with the company. Aside from the company being co-founded by a Notre Dame professor and a Notre Dame alumnus, to both of whom I was a complete stranger, it was a 100% cold contact.)

Okay, but how did I make it work logistically? That's also pretty simple. The answer is that I barely attended college after my first year, at which point I was about two-thirds of the way through my degree. I came in with 44 credits from AP tests, earned another 35 credits my first year, and then just needed 45 more credits to graduate. However, I disliked college, and being on scholarship meant there wasn't any rush to graduate super early as long as the lack of degree didn't get in my way, so I just carried on pursuing my professional goals while coasting through the rest of my degree on a bare minimum courseload, taking attendance-mandatory courses in the early morning or evening and skipping the rest to spend the day in office. (The back-of-envelope math works out: working 40h/week + attending class and doing homework 10-20h/week + tutoring 20h/week = full workload of 70-80h/week, intense but sustainable.)

(****) I call it an "obvious choice" now, but at the time, it felt like a risky bet to forego academia / big tech and take a leap of passion that most people didn't understand and advised against: "you're leaving your job as a data scientist to go... tutor?". What really helped me pull the trigger was building a bit of financial runway to mitigate the downside risk. For a couple years before taking the leap, I lived at my personal bare minimum sustinence budget while simultaneously working as much as possible. This created a safety net, maximized how long that safety net would last if I needed to use it, and got me accustomed to the long grind that awaits anyone who wishes to build their career from first principles as opposed to pattern-matching.

To be clear: when I say "bare minimum sustinence budget," I mean all-inclusive expenses sustainable for the long-term, not just Elon Musk's "$1 per day on food for a month" challenge (no hate on Elon, I just think that challenge is oversimplified to the point that achieving it doesn't properly ameliorate any financial concerns). That challenge is similar in theory, but it neglects other essential recurring expenses such as phone, insurance, rent in a sufficiently safe part of town, and gas for car (which really is essential if you need to be in numerous places throughout the day and you're not in a big city). Additionally, a single month is not long enough to know whether a budget is sustainable long-term -- for instance, at one point I cut out rent from my budget and (unnecessarily) lived out of my car, and that felt sustainable during the first month, but it turned out there was some kind of initial adrenaline / "end in sight" effect pushing me through, because after another month it felt completely unsustainable.

Admittedly, I did take this whole experiment further than I really needed to, and I'm not suggesting anyone needs to take it to the same extreme to reap the benefits. I'm just saying that the rate at which you accumulate financial freedom is proportional to your save-to-spend ratio, so if you're serious about paving your own path, then you need to also be serious about living well below your means.

(*) Here are the main experiences I had from early 2018 to early 2020, in rough chronological order:

• spending a day at Ad Astra, Elon Musk's experimental school on the SpaceX campus, which has since evolved into Astra Nova
• mentoring a math prodigy (who I continue mentoring to this day -- you may hear about him in the future)
• leading a math circle and coaching a math field day team
• developing content for math websites and even writing several of my own textbooks for fun
• tutoring & substitute teaching for Math Academy, a highly accelerated 6-12th grade math program in Pasadena where 8th graders take AP Calculus BC and high schoolers study a full undergraduate math curriculum -- the most accelerated math program in the USA
• teaching weekend test prep classes & engaging in other miscellaneous tutoring
• picking up a master's in computer science from Georgia Tech (it was only $7k, so, duh!)
• prepping a 12-year-old student for entrance to Cal State LA's Early Entrance Program (we raised his ACT math score 12 points from 16 to 28 over the course of a year and he got in)
• teaching high school at an exclusive day-and-boarding school (attendees included Usher's kids)

(**) I've been asked a couple times why I don't get back on the academic route, get a PhD, and publish on this stuff. The answer is that in terms of improving educational outcomes, science is not where the bottleneck is. The bottleneck is in practice. There are no scientific secrets -- all the important scientific principles have been understood for decades, but they are rarely used in practice (despite the best communication efforts from researchers), and up until now, nobody has put in the work to bring them all together into a full-fledged learning system.

Why not? Honestly, a better question is why at all. Developing content is incredibly slow and expensive, and building a custom application with a novel AI system from scratch requires an incredibly high level of technical skill. Why would anyone with the requisite level of skill opt to eat the opportunity cost in favor of spending a decade working like crazy, making little or negative money, in an un-sexy industry that is notorious for bureaucracy, incompetence, and shallow pockets? The level of risk is vomit-worthy.

(***) It was a huge transition point not just for Math Academy, but also for me personally. In Summer 2019 I was so close to giving up on my passion for math education that I began studying for actuarial exams -- if nothing big happened by Summer 2020, I was considering abandoning math education and becoming an actuary. The idea was to get on a well-defined career track that would lean into my mathematical strengths, pay well, and not demand too much in the way of interest, passion, or extra learning outside of math, leaving plenty of time and mental energy for hobbies.

(Why not data science? The career track in data science is not nearly as well defined, and there's a lot of extra learning outside of math: you have to know a lot more about software engineering and whatever specific domain you find yourself working in. That extra learning magnifies whatever sentiment you have about your work: it's great if you're excited about what you're working on, but it can feel soul-crushing if not.)

Even during summer 2019, it was not clear that my algorithm development work with Math Academy would grow into a big opportunity -- the work was cool, but it was ultimately just some part-time contract work on an R&D project. It wasn't until fall 2019, once we had our first real student successfully learning entirely via the algorithm without any human intervention, that I started to see glimpses of the big opportunity. But once Spring 2020 happened, the opportunity became entirely clear.