On the Health Record - Interview with Nick Sweeting, Co-Founder of Monadical & Former DrChrono Developer
We recently interviewed Nick Sweeting, a former DrChrono developer and current co-founder of the software consulting firm, Monadical, to discuss his career in programming, his time with DrChrono and what he imagines the future of healthcare technology moving towards. Listen to the full episode here.
How did you get started in programming?
It was about two years before 2014 during my junior year of high school. Our school was a one laptop per child-style high school, where everyone had to have a laptop, but the repair center in the school gave abysmal prices for fixing things like liquid damage or hard drive damage, so a few friends and I undercut the school by offering Mac repair at half the price, and the school got pretty mad at us. That was sort of our first foray into tech repair.
After that, I got a job in IT at a real company and actually dropped out of high school the second half of junior year to start programming. Then I got a job at a food delivery startup in Shanghai where they asked me in the interview, “Hey, do you know Django and Python?” And I said, “Oh yeah, I know some of that.”
I didn’t know any, so I went home that night and quickly studied up on Python and Django. Then I came in the next day, and I think they hired me because I was so cheap. I had no idea what I was doing. I’m sure I was net negative for them, but that was my first foray into Python and Django programming.
I’m sure those Python and Django programming skills were very important when you were applying for your job at DrChrono. Tell me about how you got hired here and what that hiring process looked like.
I had moved back to the US and was in Portland, Oregon to finish my senior year of high school. I think I was just reading Hacker News, and I saw a post saying, “Take this test online.” I hadn’t really done any HackerRank or interview-style processes before to apply to companies, so this seemed pretty fun. I remember it was two relatively easy Python questions. I submitted them, and then I completely forgot that I’d even applied to DrChrono. Then two weeks later, I got an email saying, “Hey, do you want to have a phone screen?”
I chatted with Daniel [Kivatinos, DrChrono’s COO and co-founder] on the phone for a while, and they offered to fly me down to the Bay area to have a week-long interview there.
Meanwhile I had finals coming up in school, so I was like, “Sure. That sounds a lot more exciting than being in school.” They flew me down to Mountain View, and they showed me around. I got to see the YC [Y Combinator] headquarters and the Googleplex, and all of that just blew my mind as a high schooler. So I was obviously wanting to impress DrChrono. I tried pretty hard during the week there to build a useful feature and ended up building the copy-paste feature in the template builder.
What kind of single focus projects were you working on at the time? Do you have a project that you were working on that was particularly memorable?
The biggest, long running project that I did for about six months was our lead generation tool for the BDR sales team. We were doing an outreach program at the time where we were trying to find every doctor in the US and work our way through contacting all of them. The directive from Michael [Nusimow, DrChrono’s CEO and co-founder] was basically, “Find me every doctor in the US. I want their phone number, I want their email, and I want their address given to the sales team.” With such a high level goal in mind, I think that’s the project I enjoyed most.
You co-founded a software consulting company. Tell us about that journey because you didn’t go straight from DrChrono into that. How did that next phase in your career come to be?
After DrChrono, I worked at a company called Mavrx that did aerial imagery of crop fields and then sold the data back to farmers. They would take the data and do a spectral image of a corn field and look for areas that are low on nitrogen based on the spectral response. Then they would feed that data into a tractor, and the tractor would drive around the field and disburse the nitrogen fertilizer where it’s needed. That was pretty fun, but I only did that for maybe six months after DrChrono before I went and started my own thing. While I was working for Mavrx, I decided to live and work remotely in Colombia, and a good friend of mine from New York, Max McCrea, went and worked there with me.
We were each working for Bay area companies and living down there, and we came up with an idea for a start-up while we were there. Max used to be a professional poker player, and we liked to play poker with our friends. We invited a lot of friends to come down to Bogotá, and we would occasionally play poker with them, but we realized that there are a lot of problems with the state of online poker right now. It’s very harsh towards beginners. They’ll join, they’ll get crushed by pro players, and they’ll have no idea what hit them. It didn’t feel equitable, and it felt like all the big companies were trying to cheat players out of their money by showing them ads. It was sort of a toxic culture.
We wanted to build a clean, ad-free site that just got rid of all that, so I ended up quitting Mavrx and joining Max in Montreal and building that for about six months. Then we raised some money in the Bay area and hired a team in Colombia. It was going pretty well, but then we went to raise our second round to get a gaming license to actually operate this site with real money. This happened during the crypto boom of 2017, and we raised money from a crypto investor. The problem was that the crypto investor only wanted to pay out our expenses each month and not the full lump sum of the investment, so he’s paying this out month by month while crypto is crashing from a thousand dollars to one for Ethereum, then down to 200 to one at the end.
So we only ended up getting a fraction of the investment money, and the company basically collapsed, and we told our team, “Do you want to stay with us, and we’ll try something else, or do you want to go off and work for other companies?” We really liked the team, and I think they liked us too because they decided to stay. That’s when we switched to consulting. Then for the last two years we’ve been building this consulting business [Monadical], and it’s been going pretty well.
I have a question here from a DrChrono employee that isn’t me. First question - if I’m looking for custom development work, what’s the best way to express my idea to make it a tech reality?
I think honestly the best way is creating a lot of drawings. There’s nothing greater than a pencil and paper or a whiteboard when it comes to expressing a complicated idea to someone else. You can spend an hour talking with someone just going around in circles. No one can really visualize the same thing unless you have a shared visualization that you can then both sketch around and mock up.
As software consultants ourselves, the majority of our sales process is saying no to clients. When we hop on the phone, all of our clients will gush about, “Oh, we want this feature and this feature and this feature,” and they get to the end, and we ask for their budget. Then they give the number and we’re like, “Okay, so we’re going to say no to this, this, this, this, this, and this.” Then there’s one thing left, and that’s your core product.
A major piece of building successful projects is saying no to things. [As a client] you can do that in advance by making a big pile of features that you could conceivably want. Then rank order them, then cross off 90% of them, and then come to us with that list. That’s probably the most effective way to spec out a project.
How do you think cutting edge technologies will change healthcare for doctors in the next 10 years?
I think a lot of the menial parts of healthcare will be able to be automated. There’s this general conception from people outside of healthcare that doctors are the end-all be-all of medical advice. You go to a doctor, they give you the answer, it’s the right answer, and you don’t question it. That’s probably true for the vast majority of cases, but that’s not necessarily true always. I think we’re going to see a rise of technologies that sort of prove that to not be the case, whether it’s machine learning being trained on data sets of tens of thousands of people to learn what diagnoses were correct, or whether it’s an automated system that flags when the wrong medication is prescribed by accident.
We’re going to see a lot of cases where tech will provide a correct answer when a doctor gave a wrong answer, or maybe a doctor gives a right answer when tech gives a wrong answer. We’ll see more and more clashes between automated care advice and human driven care advice, and managing that on the tech side is going to be really interesting.
So I imagine that you’re talking about AI and combing through data to recommend a solution, right? It’s not like we can’t do the machine learning algorithms. We know how the science works. It’s just a matter of how we get it into the doctor’s office, so that when they’re visiting a patient, we can generate that on the fly. Where are the big challenges do you think?
I still think the biggest challenge is ethics. We have technology that’s mostly correct, and that’s a very dangerous thing. Because if you just release these systems into the wild, people are going to die, people are going to hate the technology, all sorts of bad externalities will happen. I think by far the biggest challenge is doing the ethics research to understand what’s actually going on inside of these models when they provide a recommendation. We don’t have tooling to really see why a model is making the decision that it’s making. Building that tooling to be able to introspect how a neural net works is a big part of doing AI these days.
Another part is understanding biases. Often neural nets will be trained on a dataset that has a lot of inherent bias in it. When you try to give equitable care to people, that bias comes out in the advice that the neural net gives. So whether it’s giving different advice depending on someone’s race or gender, there’s just a lot of gnarly dragons in there that need to be resolved before we just release these AI systems into the wild and tell them to treat people.
Listen and subscribe to the podcast here.