I'm excited to share that I am officially a founder again!
This time, I'm building something close to heart: not a healthcare SaaS platform, but a tool that helps developers find and fix production issues 10x faster.
As I embark on this adventure, I want to take a moment to reflect on what led me to this point and why building a dev tool makes sense for me.
~15 years ago, I learned how to code. It changed my life for the better. The part I enjoyed the most was the freedom of expression; the ability to create anything you could imagine.
Over time I found myself drawn more towards the process than the outcome. I got curious about the craft of software development and how software is "delivered" to "production". That curiosity evolved into a decade-plus-long obsession with DevOps, developer experience, and dev tools.
It might sound like a strange thing to be passionate about. Most devs consider stuff like integration tests, CI/CD, and preview environments to be annoying. Necessary evils. Barriers to shipping "real customer value". But for me, these are the interesting parts.
I'm 31 now with over ten years of professional software engineering experience. As I look back on the first major arc of my career, I can identify four "chapters" that shaped my predisposition towards dev tools.
Chapter 1: Dev Bootcamp.
In 2014, while still studying in college, I enrolled in Dev Bootcamp, one of the first in-person coding bootcamps. A North Carolina native, I wasn’t exactly surrounded by tech and startups during my adolescence. I discovered DBC through a Forbes article and pitched my parents on attending over the summer (it was that or an investment banking internship).
DBC taught Ruby on Rails, one of the leading web frameworks at the time. I distinctly remember the "wow" factor of Rails. Ruby code was easy to read, the Rails framework was really thoughtfully constructed, and everything seemed to "just work" out-of-the-box. In retrospect, this was the first time I experienced good developer experience. It left a big impression on me.
Chapter 2: Crunchbase.
In 2017, I moved west to San Francisco and joined Crunchbase as a "Frontend Engineer in Test" (a pretty pathetic title). I didn't do much frontend engineering. Instead, I implemented integration tests and CI/CD. This led me to the Infra team, where I found a much better home.
Crunchbase is where I learned that quality, testing, deployments, and reliability were serious challenges for modern engineering teams and fascinating problems of their own. It's also where I realized that infrastructure engineers are frequently the most technical, talented people within their organizations.
Chapter 3: Brex.
In 2019, I joined Brex as employee ~70, right as the company was entering hyper-growth.
As a young person getting started in Silicon Valley, everyone tells you to "join a rocketship". Fast-growing companies are the best places to learn. Turns out, it's true.
At Brex, I had the opportunity to start not one, but three distinct engineering teams during my tenure: Infrastructure, Observability, and Bill Pay (a product team). I also got to hire and onboard dozens - maybe hundreds? - of engineers, which meant thinking a LOT about developer experience and internal tools.
To be honest, I was completely out of my depth during my first year at Brex. But the experience made me “cracked” and showed me what a high-performing engineering org looks like, which was invaluable. I’m lucky to have worked there.
Chapter 4: Opkit.
In 2021, I was bored at Brex and decided to start a company.
Opkit was a medical billing startup that evolved into a healthcare voice AI company following the launch of ChatGPT. For 3.5 years, my co-founder Justin and I built software for medical practices, hospitals, and telehealth companies. It was a challenging, rewarding, and most of all humbling experience.
People often ask why I started Opkit. I wasn't a healthcare expert. In fact, I had no healthcare background whatsoever. Justin and I seemed like a pretty bad fit for our industry.
The official answer has two components...
First, while working at Brex, I developed a thesis about verticalized fintech. 2021 was both the peak of a fintech wave and the middle of the covid pandemic. It seemed obvious that fintech was going to have a massive impact on this particular area of the economy.
I also had an unfair advantage in healthcare: my dad was a practicing orthopedic surgeon. Through him, I was able to interview medical billers, tour facilities, and pitch hospital executives.
So that's it. The official answer. And while it satisfies most people, the reality is that there's more to the story. Truth is, I started Opkit by accident.
Ever since DBC, my dream was to start a venture-backed tech startup. My cohort-mates and I read Hacker News religiously. We idolized founders like the Collison brothers. For years afterwards, even while happily working as an engineer at Crunchbase and Brex, I made secret plans to one day go off on my own.
So frankly I didn't care what the idea was. I just wanted to START SOMETHING. I never thought deeply about whether healthcare was right for me.
When we got admitted to Y Combinator's summer 2021 batch, I was legitimately surprised; I didn't think they would go for the idea. I couldn't bring myself to turn YC down.
It took another year for me to discover that healthcare back office is NOT my passion, and a few more years after that to realize that passion is a prerequisite for success. These were tough, expensive lessons.
Nevertheless, Opkit found a lot of success: we raised a few million in VC funding, had multiple big launches, got featured by Axios, and built/commercialized one of the first LLM-based voice agents.
Eventually, when faced with the decision of raising additional funding vs. seeking acquisition, we chose the latter. In September 2024, Justin and I sold Opkit to 11x, a series A startup in the go-to-market space. We landed the plane!
11x was a roller coaster ride: at times chaotic, but also exciting and rewarding. The [ex-]Opkit team was tasked with rebuilding Alice, 11x's AI sales rep, from the ground up using then-novel agentic patterns and technologies. We ended up building one of the first and largest agents on LangChain.
I'm grateful to Hasan, Prabhav, and the 11x team for the opportunity to build Alice 2 and for bringing me back to San Francisco at the height of the AI revolution. There's truly no place I would rather be right now.
~8 months into 11x, the Alice rebuild was essentially complete. At this point, I had left my old life in New York completely behind. I was now consumed with thoughts about AI agents.
That's when I had my lightbulb moment.
One day, while investigating a bug report, I was struck by a contradiction: here I was, Tech Lead for a futuristic AI product, which my team and I had built using futuristic AI coding tools...
And I’m debugging it with the same set of observability tools that I’ve used for my entire career!
Datadog. Sentry. AWS CloudWatch. After years of using them, their interfaces had become second nature. I rarely gave them a second thought.
But on this particular day in summer 2024, when it seemed like EVERYTHING about my job was changing faster than I could blink, I was surprised to notice that these tools were - essentially - the same.
From there, I started to get angry. It shouldn't take hours of digging through Datadog to debug a production issue. It shouldn't require a week of engineering work to properly set up distributed tracing. I shouldn't have to create, manage, and manually triage all of these damn monitors!
I started imagining ways AI could be used to improve this part of the software development process, drawing on whatever "taste" I had accumulated from years of thinking about developers as customers. Soon, I became completely obsessed with this problem and opportunity. I decided to leave 11x to start working on it full-time.
If you've made it this far, you're probably ready for some specifics about what I'm building today. Unfortunately, I don't have much to share right now. Careful readers can probably discern the shape from details in this post - or by following my Twitter feed. I promise to share more soon. :)
As I wrap up this reflection, I want to thank the MANY people who have supported me over the years, including but definitely not limited to:
- Aaron Epstein
- Amy Xiao
- Andy Pham
- Ben Baltes
- Christine Keung
- Cos Nicolaescu
- Dalton Caldwell
- Harrison Chase
- Hasan Sukkar
- Jarrod Ruhland
- Justin Appler
- Justin Ko
- Lindsay Pettingill
- Merrill Lutsky
- Neal Hardesty
- Prabhav Jain
- Rafa Corral
- Reshma Khilnani
- Rex Salisbury
- Robert Conrad
- Thomas Césaré-Herriau
- Vamsi Chitters
(Alphabetical order.)
Anyways, here's to round two!