Why does another Mass times site exist? Read my story →
Winning a hackathon gassed me up in the wrong way. I built a startup, choked, tried therapy, then built a local AI that remembers 28 years of my life.
Loading...
Winning a major hackathon didn't ruin my career, but it definitely gassed me up in the wrong way.
I was already working in startups. I knew the drill. But that win planted a specific kind of delusion. It tricked me into thinking I was the main character. It convinced me that I was done being a soldier and was ready to be a King - VC funding, IPO, the whole "Founder" mythology.
I quit my job to build a startup. This was back when chatbots were text-based NLP, not the AI giants of today. I built an SMS notification platform. The engineering was solid. I did the SEO research, scraped pages of competitor leads, and had a list of potential customers ready to call.
And then I choked.
I remember staring at that list of phone numbers, paralyzed. I would "polish" the interface for the tenth time just to feel busy. I would step outside just to avoid the screen waiting inside. I realized, with a sinking feeling, that I had the arrogance of a founder but none of the humility. I failed because I was too proud to do the grunt work of selling - or at least to pick up the phone and find out if I sucked at it.
I went through a few therapists during that dark period. While they were helpful in many ways, I kept running into a common friction point: the gap in shared context.
It is incredibly difficult to find a provider who sits at the specific intersection of being an immigrant, a first-born Chinese son, and a tech founder. I found myself spending expensive hours just trying to explain the baseline reality of my life - the specific pressure I put on myself to "make it" in a new country, or why the standard advice about "work-life balance" felt wrong. Working blindly hard was the only thing that had ever gotten me ahead; being told to stop felt like being told to surrender.
The AI didn't slip in until much later, long after those therapy sessions had ended. When I revisited the memory of my failed startup, it didn't ask me to explain the mechanics - it just pulled up data on the commoditization of that market and the race to the bottom in that specific vertical.
It intellectualized my pain, which is exactly what I wanted. It normalized the failure through data, not just empathy.
But as I dug deeper, I noticed the "Memory" features getting stronger. I was feeding years of personal context into a black box I didn't own.
I realized I could build something with that same long-term memory - but privately.
So I built something.
In early 2025, I put together Project Trace. I fed it 28 years of my digital life - journals, emails, notes, everything I'd hoarded since 1997. Over 10,000 fragments.
It wasn't just storage. I could ask it why I kept quitting, and it would surface entries I'd forgotten - connecting dots across decades that I couldn't see myself.
I'd built a second brain that remembered everything I'd tried to forget. (If you want the technical details, here's the architecture.)
I was proud enough to show it to a close family member. I sat her down and explained the tool. I talked about how we are the guardians of our own memories, and how this tool could let us preserve our stories for the generations that come after us.
I thought she would appreciate the duty of it.
She didn't look impressed.
She watched the tool surface these memories - the struggles, the pain, the things we usually keep hidden - and she recoiled. She didn't see a "legacy." She saw someone holding onto pain that should have been let go.
I put Project Trace on the shelf.
Turns out building a perfect mirror of your life doesn't mean you want to stare into it every day. Maybe the problem was that I built it for myself. A sample size of one.
But it taught me something. The data matters. The raw signal matters. And maybe there's a way to preserve it that isn't about looking back - but about leaving something behind.
I finally dusted it off. Here's Project Trace in action - a local AI that remembers 28 years of my life:
This is the second in a series about my return to engineering. If you want the technical details, here's the architecture.