My journey begins in sophomore year of high school when I started to develop an interest in airsoft. If you’re not familiar, it is similar to paintball but allows for much more realistic simulations because of the decreased projectile size and mass. I didn’t have a field nearby to play very often, so spent most of my time tinkering and improving. One of the obvious bottlenecks were the electrical components, which consisted of 20awg wire and a physical switch that were expected to carry 30+ amps continuously with some spikes into the 100s. At that point, I knew a little bit about electronics from being on my school’s robotics team, so decided to come up with a better solution using a MOSFET. After experimenting for a few weeks, I came up with a design that worked and posted about it online. In addition to the safety benefits, my device would improve the rate of fire by sometimes up to 20%, a very large improvement for something that only cost me about $5 to make.
I was surprised a few hours later when someone messaged me wanting to purchase one. At the time, I didn’t quite understand why since the tutorial I wrote was freely available, but I had spare parts available. He sent me money on PayPal and I shipped it out the next day. Over the next two and a half years, that original device (which people took to calling a MOSFET) went through three major design changes and was shipped out to over one thousand customers with a satisfaction rate of above 95%. This business, which I called ZC Airsoft, would let me get away with not getting a “real” job all throughout high school and would also set the stage for my later interest in computer science. You can read more about all of my (mis)adventures in starting ZC Airsoft in my blog post here, which also contains everything I would do differently if I knew everything that I do now.
The really interesting ramification of using a MOSFET instead of a physical switch was the possibility for programming custom functionality with a microcontroller. But there was one thing stopping me from making this: I absolutely hated programming. [ ]
I applied to all the top technical schools and got rejected from every single one of them, only getting in to my safety school: Rensselaer Polytechnic Institute. My mom attributed this to admissions officers being biased against Asian males, which is somewhat true, but I think a more accurate story is that I had middling grades, no sports, and pretty badly written essays with the only saving grace being a perfect ACT score.
Anyway, in the summer before starting college, ZC Airsoft was for the most part making money without much effort needed by me (by that time I had outsourced both manufacturing and fulfillment). To pass the time, I decided to read one book a day. I consumed books in all different sorts of genres; one day I would read Richard Feynman’s autobiography and the next day read The Lean Startup, or Code Complete 2, or The Power of Full Engagement. In retrospect, this was perhaps the best decision I’ve ever made. I learned more practical things during this summer than everything I had ever learned in school. My love of reading continues to this day and I attribute any success I have or will have to the books I have read.
During my first semester at RPI, I switched my major seven times, starting from Nuclear Engineering and ending at Computer Science and Mathematics, which seemed like the best optimization between what I was interested in and what was most in demand. In my second semester, I was brought on as an Inventor to The Inventor’s Guild, where I made some money contracting for a few local startups, attended my first hackathons, and tried to commercialize the work I had previously done in sports analytics (though unfortunately failed).
That summer, I got a job as a Data Science Intern at Haystax Technologies in Washington D.C. [ ]
[sophomore year: guild ceo, deep remix, contrary capital, case urp, venue, linux]
And that brings us to today! I am currently working as a Deep Learning Intern in Workday’s SYMAN data science team to develop deep neural networks for automated expense reporting. More specifically, I am designing and training a state-of-the-art fully-convolutional regression network for text bounding box detection. I don’t have any results to report just yet, but I will update this page when I do!
My current side project is [Tech Intern Blueprint]
This Fall, I will be working for Amazon as a Software Development Engineer Intern in the [ ] team.
Looking further ahead, I’m trying to get more experience in deep learning and in the software engineering around it. Specifically, I am looking for internships and co-ops in the Spring and Summer of 2018 where I can add massive value to the company while gaining this experience. After finishing school, I plan to work in or around deep learning at a fast-growing early-stage startup in the Bay Area. My eventual goal is to use my skills and experience to found a company that can put a dent in the universe.
I am always looking to get connect with interesting people! If you are in San Francisco and want to get coffee, email me at firstname.lastname@example.org
Last update was 2017-08-01.