While writing this, I thought: 2023 wasn’t really anything special. After writing this out, I think it was the most intense impactful year in my life.
Let’s do a recap:
January
On January 10th, I had a call with one of my friends at 3AM. The next day, I caught up with another friend.
We ended up talking about the same shit.
We all realized:
- we did not want to work for the rest of our lives
- we all wanted financial freedom for the same reasons: giving back to family, comfortable living, generational wealth for your kids and their kids, and to live like a fucking G when you’re young in the one life you’re given
- and that we were all fucking brokies
Snippets of the conversations stood out to me so much I even wrote it down on a GDoc I titled “3am manifesto”
Money was a big issue in my household growing up.
This HAD to change.
The 3 of us founded a Discord accountability group to keep ourselves accountable throughout this year. We called it the War Room. The criteria for membership was: seek relentless self improvement, don’t compromise on ambition. Referral based entry only.
This group was the best decision I ever made.
The next day I ideated and started working on business ideas. It was time to thread the needle.
The first was a SMMA/Digital Marketing Agency, called Nero Solutions. A friend and I worked on this together. Our business' service was providing small businesses with online marketing services and website building.
The Ontario government was running a grant program for SMB’s as well, which they could apply for and receive free money to invest into their digital marketing efforts.
The pitch would be: if you haven’t taken advantage of this grant yet, do it, and we will take a portion of it to build a new website or do a makeover for your existing one. The money would come out of the government’s pocket, and not their own.
We made a shortlist of potential client phone numbers by literally panning Google Maps for hours and finding SMB’s that didn’t have a website. The next step was working on a sales pitch script, and start cold calling like crazy.
For the next week or two, we prepared website templates, studied landing pages, and cold called. But, my parter got an enticing engineering job offer at a startup and didn’t have time to continue with the agency work. Being the pussy I was, I stopped working on it as well. Didn’t feel confident in doing this venture on my own.
Feb-March
I quickly pursued another business idea: automated tabular data extraction from PDF/non-tabular file formats. There was a much bigger existing demand for this service in the market. Accountants and researchers often need to pull tabular data from PDFs into Excel/Sheets to do calculations. But, this can be a pretty time consuming process to do by hand, since tables (PDFs in general) aren’t copy pastable. I could save hundreds of hours of processing work by building something to automate this.
I named the project Tabulate AI. I cold emailed 20 accounting firms and 2 got back to me. I pitched the idea of an automated software they could use, similar to snipping tool’s UX. They were interested in being my first clients/giving feedback as I developed it.
I prototyped an OCR-based proof of concept by end of Feb. I did the engineering work with my younger brother.
I turned 20 while building out Tabulate. The urgency was amplified. I had to make it.
At this point momentum was strong. Besides myself, my other friend in the War Room who was building his business was making good progress. Daily accountability check-ins were awesome.
Unfortunately, the technical capabilities were just not there yet in any open source OCR models to build a robust error-free pipeline from image -> tabular data. Reminder to readers that at this time GPT-4 was not even out. So, Tabulate went on the shelf, and I went back to the drawing board.
March-April
At this point, I had other items consuming my bandwidth since before the year:
- learning how to do full stack dev properly (learning React and JS deeply)
- doing sponsorship outreach for CUTC and organizing a conference
- starting a friday night reading group for ML papers with a close friend, and also planning on a summer robotics project
- going to several ML conferences
- continuing my existing part time ML research commitment during school, building an eval framework for evaluating DL models on dynamical systems
My next idea was a weekly news digest service. It would scrape news articles every day, provide news digests using GPT as a summarizer tool. It would be monthly subscription based. Users could indicate preference for the news sites/types of news they want to see, with an eventual goal of a generic webscraper that could be used for any blog/jorunal/news source. This one was called “Digestibles”.
Unfortunately, midterms and exam season came swinging in like a one-two punch. After neglecting school classes for weeks, I was pretty packed in terms of bandwidth. So Digestibles never got the fair shot I should’ve given it.
Before I knew it, the term was over. Juggling these “side hustles” with school was more difficult than I thought. It was very hard to balance and compartmentalize, usually something I’m really good at. But even as I’m writing this, I didn’t realize just how many commitments I was involved in. I was spread pretty thin.
On school: it was my 2nd term in the Computer Engineering stream (after making the transfer out of EngSci). I learned so much about computer architecture, low level logic, and electronics. I even built a simplified GIS Application (essentially Google Maps) from scratch, writing thousands of lines of code and doing really neat low level optimizations for path algorithms and reducing rendering latency.
It was also the first time school also didn’t feel as easy. Concepts were harder, so I had to take more time to understand them.
May
Once exam season ended, I started working on Digestibles again. Started having co-working calls in the War Room to keep myself accountable. But it was hard to bring back life into it.
For one, I was really burnt out at this point. Cramming for exams, the mental strain of writing them, I caught COVID (again), and the CUTC conference had just happened. I needed a recovery break.
Secondly, I started my summer research job and a robotics project, which took away most of my time.
On summer job: It unfortunately did not pan out as expected. The pay I realized was a lot lower than what I thought, the project was different and not directly ML related, and I was in general disillusioned with academia at this point. I realized that I wasn’t going to achieve financial freedom in academia. It was the wrong crowd.
Moreover, the work wasn’t fulfilling. I was given way too much creative freedom over my project. I was like a PhD student, designing my own experiments, hypotheses, and doing the entire science myself. This was great for PhD students who wanted to do research in their deep interests. Not great for an undergrad who wanted to do exploratory research/was only slightly curious.
Because of all above, I was in a pretty depressed state in May. Moreover, some back of the envelope calculations made the pricing of Digestibles near impossible to bootstrap.
I put a close on Digestibles. At this point my family was going through some medical struggles as well which made it worse.
I was in a lull. I knew what I wanted (financial freedom). But 3 failed attempts later at building an income generating side project, things were looking bleak.
June
I spent time travelling, reading, and tinkering on non-finance driven projects like an autonomous electrochemistry robot. Caught up with old friends and faces at SF Tech Week and the 1517 Summit.
At tech week, my friend and I LARP’d as co-founders of this robotics company “Open Science” working on affordable autonomous self-driving labs. Stretching the truth, but not totally lying. We had results from our project work during May-June, and a prototype demo video. We used this to get into founder-only networing events and VC mixers. It was a fun time making the pitch during the Uber ride to the events and completely improvising. Talk about fake it till you make it lol.
There are thoughts I have about the whole VC/founder dynamic in SF, but that’s for another day.
July
At this point, I also became really interested in neuroscience. Jeff Hawkin’s A Thousand Brains book really shifted my perspective on modern day neural nets and their limitations. I wanted to tinker with graph based data structures that would replicate intra-neuron level circuits, not just the reductionist output = weighted sum of inputs model. I got the 1517 Medici grant to pursue this idea.
I also got introduced to a prof at MIT through one of my friends. He, my friend and I started working on an adjacent reserach project together: seeing how different forms of KGs used in KG based RAG would affect LLM performance. If we found that hierarchal based KG construction worked better than non-hierarchal methods, that would mean hierarchy is an important attribute for this “evolved memory system” we were trying to build.
Halfway through the project, I was given the offer to join as the founding engineer at MyMi, a startup that would be based around this technology. We would coin it as Mirrored Intelligence, a form of AI that would take on the personality and knowledge base of any given set of data.
I eventually took the offer. Few reasons:
- After 3rd year in school, my plan for my co-op year was to find a job at an early stage startup to learn a lot of technical skills and get my hands messy. This would just be the job coming 1 year earlier than initially planned. I also saw this as a potential exit strategy from school early, in the chance of a miracle exit early on.
- It was in a great area to be in given the AI hype, and I’d be making a lot more money than I would be in research.
- I liked the founder. Having worked on a research project already, I thought we had good chemistry and he had a good head on his shoulders.
August
Early musings of MyMi.
At this point the War Room grew into 7 member group.
One of my good friends was visiting Toronto during this time (who also was a founding member of it), so we had our first little official meetup 4/7 present.
For the first time there was clarity, semblance of balanced routine, and a healthy balance between rewarding work and non-work time. I picked up piano again, caught up with a ton of friends, and built my first ever PC rig. Was learning a ton on the job and growing as well.
Gap year had begun.
September-November
Heads down in monk mode. Moved to Silicon Valley. Moving took its tolls:
It was a gureling period of hard work. We had important demos to showcase, investor meetings to raise a seed, and made numerous minor pivots to our strategies. This was the highest growth period, technically and learning about entrepreneurship, in my life.
But being in person meant I was living pretty much in isolation spending 10hrs+ every day in the office. There was no room in the day for other things in my life. I neglected regular gym, my diet was shit (I gained weight), and lost a bit of hair.
Although I got better at setting boundaries and balancing my life a bit, it was still a mental strain living alone in a new city.
I found solace in music and the support and motivation seeing other War Room members grow. Nothing good comes easy. I embraced the stress as a good thing.
During this time I shipped a fuck ton of code: (does not include library/node modules)
Frontend Backend
November
I was back in Toronto for a few weeks for family reasons. I continued work remotely, but at home. I was mentally in a better state and my health was recovering. My hair came back.
I was noticably burnt out from the intense work sprints from weeks prior. Daytimes felt sluggish. It could fully be that it’s just seasonal depression caused by Toronto winters, but looking back I think 80hr+ weeks were not a sustainable routine for me. It wasn’t just the hours, I was horribly inefficient. There were days where I got more done in 2 hrs than the previous 10.
But questions started formulating in my head.
Was I on the right path? My friend suggested to think about things from a first principles perspective. I asked myself: What do I really want in life? What is success to me? Why am I doing this?
The answer since January had always been financial freedom.
But nowhere in those two words does it hint having to achieve it through a hardcore startup.
There were many dead remote SWE jobs I could just rack up, like another friend was doing (working 2 easy jobs at once, making good income). I learned a lot of lessons and felt confident in building smaller scale side projects. Or freelancing like my other friend, who was doing over 6 figures a year.
Hell, why even limit myself to tech?
I returned back to Palo Alto with these thoughts over my head, preparing for another season in the hyperbolic time chamber.
December
This time I made conscious effort to keep a healthier balance. I went to the gym in the mornings, and did shorter office sessions but in more concentrated efforts.
I didn’t notice any drop in performance, and we were on time delivering all deliverables.
I was a lot happier. I return back to Toronto for the holidays, spending lots of time with family and catching up with friends.
Thoughts going into 2024
I’m revisiting those questions now:
What do I really want in life? What is success to me? Why am I doing this?
There are clearly “easier” options to amass wealth.
Yet I don’t do any of these “easier” lower risk plays. Why?? Why the hard path?? Logically it makes zero sense. The amount of hardships I’ve seen that come from building a startup is indescribable. So much solitude and sacrifice.
Even Jensen (NVIDIA) said he would not start NVIDIA again had he known how much work it was.
But more importantly: it’s a hard pill for me to swallow, but I am not addicted to software engineering. After getting a taste of it, I cannot see myself writing code as a full time job for more than 4 years. Even 2. I genuinely can’t. I never once saw it as an end to a means, but always as a means to an end. A tool to use here and there to do what I want.
Maybe I hate the idea of copping out to something easy. But that wouldn’t make sense, since I was candid dropping into a less prestegious engineering program. Maybe I’m just unhealthily seducted by the tech industry? Is this stockholm syndrome??
Or maybe I’m scared of re-aligning my goals for this gap year, having sunk-cost fallacy and afraid of feeling like the last few months were a waste. Maybe I’m concerned I’ll burn invaluable bridges with the current team if I diverged onto my own path at this moment. I’m not sure.
I’ve reflected and found three things I would genuinely say I’m “addicted to”, where I could spend countless hours working on and not get burnt out in any capacity.
Analysis: analysing arguments, logic, optimizing for efficiency, identifying flaws, etc. Reflected in my love for strategy games since I was young, I love optimizing the shit out of something to get the best results.
Music: performing, consuming, learning about music history, or discovering sick remixes or mashups. Growing up, music was a significant part of my life – I learned how to play many instruments. I fucking loved it.
Psychology: my neuroscience curiosity is stronly driven by a curiosity behind human psychology. At the end of the day there’s a common shared monkey brain in all of us. To an extent we are victims of manipulation to forces around us. Exploring those connections: whether it’s social media algorithms, mimetic behavior/social contagions, understanding charisma/body language, or negotiating/sales tactics, it all falls under the hood of human psychology.
2024 will be the year I make it. 2023 curated and clarifyed the drive behind why I do the things I do.
And it’s given me a taste of what the real world is like in the Silicon Valley/tech world lifestyle that I only saw through the Internet. A harsh reality check.
The north star is clear: financial freedom.
I’ve tried many things, I am going to try many more. I think ecommerce fits right up the alley in the intersection of all 3 of my passions. We’ll start there, see where we end up.
See you next year.