This interview was originally published in the September 2013 issue of Frankly Speaking, the unofficial newspaper of Olin College. It has been modified to make it more accessible to the general public.
A Candid Conversation with Jialiya Huang about the changing face of hardware development, working with co-founders, and what it feels like to get a company off the ground.
Jialiya Huang, Olin class of 2013.5 [ed. note: the ’.5’ denotes a December graduation], founded Technical Machine with Tim Ryan (Olin ‘13.5) and Jon McKay (Olin ‘13) this summer. The company is launching Tessel, their first product, on September 5th, and is both thrilled and innervated by all the interest the Tessel has received already on Hacker News, Hackaday, and Japanese Slashdot.
Full disclosure, I’m working for Technical Machine too– mostly on press and marketing at the moment [Kelsey Breseman]. But it was still a great opportunity to speak with Jia at length about the future of hardware development and her personal goals in starting a company.
BRESEMAN: What is the mission of Technical Machine?
HUANG: Technical Machine is trying to make it easy for software developers to get into hardware. It’s a different take on our last year’s SCOPE project. [SCOPE is Olin College’s senior capstone in engineering.]
Hardware is something that I’ve always wanted to get into, but never felt like I was ready to, because it always seemed like there was a lower level that I needed to get knowledge of.
For something like web development, I always knew that there was a tutorial, and I could make it, and boom, there’s a site. That’s it.
As I got into hardware, it was like, here’s a simple circuit that can flip an LED on and off, on a 5-5 timer, for example. And then now, well, you can do that with transistors instead of just a 5-5 chip.
Well, crap, now I need to know about transistors. You can just keep on going lower and lower, and I wasn’t sure where it stopped.
That always has been a fear of mine, because I was never that into hardware. Technical Machine is trying to make that problem easier.
BRESEMAN: What do you want from hardware?
HUANG: I very much see everything as– that should behave like software. There should be a well-established API call for any action that I want to accomplish, or I should be able to look at how other people have done it and be able to replicate their actions.
With the open hardware movement, it’s getting easier, but back in highschool, I never knew that was a thing. I wouldn’t even know where to start first. Even when I was taking circuits classes, learning physics, all the stuff that we did was very much textbook-like. It wasn’t me designing my own circuits. Very much so when I started computer science, it got to– oh look, here is a simple program, and then I was modifying it, I was adding to it, I could make it into whatever I wanted it to do. The steps just really weren’t there with hardware.
Technical Machine, and the Tessel in particular, is a way of trying to expose an entire world to software developers, who otherwise wouldn’t have gotten into it.
BRESEMAN: Is Tessel just a starting place for your users to get into hardware?
HUANG: Right now, when software developers look at something like hardware, the first thing that I think of is, if I create something cool, how do I share it with the world? That’s one of the biggest thing that I want to do with any software library that I write, is put it up in the open, have people using it, and put my name out there.
With hardware, it’s just like, this is cool, I can put that in my room; all my friends will know about it, but that will never get out into the world.
Tessel is trying to make it easier for that entire process to happen.
The set of knowledge that you would get from making one hardware device doesn’t really translate into making more of them. Tessel is just trying to streamline that entire process so that if you can make one, you can make ten; there’s a service that lets you make 100, or make 1,000.
BRESEMAN: How does Tessel streamline that process?
HUANG: Some of the things that we looked into in particular were other services: an API for hardware, essentially. You have all these capabilities that you want on your device– let’s say I want Bluetooth and accelerometer, for remote sensing or something. And that’s all I need. And I can do that on an Arduino: I can get an Arduino accelerometer shield; I can get an Arduino Bluetooth shield. But in order to migrate onto my own PCB requires me to now know hardware design. Whereas with the Tessel, what I hope to accomplish is making a service so that someone can say, here’s what I need, and then we send them back something that is exactly that.
BRESEMAN: Did you make Tessel to start a company, or did you start a company to make Tessel?
HUANG: A bit of both. Somewhere along, sophomore, junior year, Jon and Tim and I decided that we should be doing a company. At that point, I think it was the summer after sophomore year that I was at Amazon and Jon and Tim were at Microsoft, and we realized what cushy jobs we had.
It scared me that I could see myself doing something like that. It scared me that I could be static, just spend the next 10-15 years of my life doing nothing in particular, and yet still getting paid. It was weird that I was doing less work than when I was at school, and yet I was making a lot more money than I ever had in my life before.
I think that both Jon and Tim felt the same unease with that situation. We decided, the only way out of this is to get way out of this and do our own thing. And then we were trying to think of products and things we could make.
Along the years, there’s been half-baked ideas that we keep pushing out. They’ve all been very much restricted to software only. It wasn’t until SCOPE that we had much hardware experience at all, and then we just suddenly launched ourselves into hardware, and we just found so many new issues that we could solve.
The problem with software is that it’s so crowded is that anything that we would have experience with students is catering to the educational market, which is kind of a bad market to be in, or like, a hobbyist playful, nothing really necessary market, because we don’t really have domain expertise in anything else. However, one of the main things that we do have skill in is software, and if we can take the software paradigm and establish it in hardware, that’s one of the things that we’re trying to do now.
BRESEMAN: What software paradigm are you bringing into hardware?
HUANG: It starts with the community. That’s what the open hardware movement is trying to do. Right now, if I had a question on Rails programming, no matter how weird it is, a weird error message, I could just Google it, and then up on StackOverflow, there would probably be someone else asking the exact same thing. Something like that is so standardized, and so many people are using it, that there’s very little risk in putting your skill set in there and putting your time in there and making it even better.
However in hardware, the space is kind of fragmented in that sense. If you get really good at one particular microchip and all of a sudden you’re moving to another one. Now you’re learning an entirely new chipset, you’re learning new instructions. Your knowledge isn’t that transferrable. So whatever you put into one hardware set, you’re stuck there, in a sense. That prevents people from going out and diving really deeply into these technologies, because there’s no easy fallback.
If you were the only person doing a startup in the world, and your startup fails, no one else would hire you. But if you’re doing a startup and there’s ten other companies in your related space, one of them will probably snap you up for your domain expertise alone.
BRESEMAN: It’s not like that in hardware?
HUANG: In hardware, I’m actually not sure about this, because I’m not a real electrical engineer. It is definitely a lot harder to, one, formulate hardware questions about hardware such that it is easily searchable, so that could just be my bad, but two, even when we do run into hardware issues, the way you find- especially for the NXP chip that we’re currently using– I have an issue about how to configure SDRAM, for example. I look it up and the top replies are one guy, on a forum post. This is literally like ten years ago for software, how these things are answered: weird forums in the middle of the internet somewhere.
BRESEMAN: How does Tessel solve that?
HUANG: With Tessel the idea is that our layer would abstract away those kind of questions so that you wouldn’t need to ask those questions until you’re at a point where you’ve proven out your product and you have enough money to hire someone to do that.
When you’re first going into the hardware space, the number one thing that you want to do is establish that your hardware product would be successful if you launched it. Getting that as quickly as possible reduces your risk in entering that space.
BRESEMAN: What is your ideal market?
That’s great for starting out in the hobbyist space, that’s great for building up a community, and that’s great for us figuring out what the overall market would want. And there’s also the space of enterprise-y things: it would be cool if my washing machine had an API that I could interact with, or it would be cool if my oven did, or my everyday devices. I wouldn’t have to open up my washer, figure out what chipset it’s using, and there are ways to hack around programming every single chip– if there was a way for me to easily control physical devices out there. That’s a different enterprise market that we’re trying to breach, but we have to gain legitimacy in the hobbyist market before we can go into that.
BRESEMAN: What is most exciting to you about Technical Machine?
HUANG: Everything is incredibly exciting because I have no idea what’s going to come the next day.
For the longest time, we were unsure about technical development, the first few months. That was a lot of getting our heads down, making sure that what we were proposing was feasible. Of course, that has its own highs and lows. And then all of a sudden we launched on Hacker News, and suddenly we were getting so much attention that we had to respond to it, and it’s just been building up.
There are so many decisions that we need to take care of, and that we need to think about. It’s very different from any other job I’ve had. I’ve been an intern at other places, but those are very much low-level. You get thrown a very specific task, or a chunk of a task; the parameters are very well-defined. But with Technical Machine, because we’re creating this company and creating the market as we go along. Every decision that we make is impactful.
BRESEMAN: What are you worried about at the moment?
HUANG: Absolutely everything. I’m worried about moving out of my house, I’m worried about finding enough things to eat. What happens when we’re no longer at Highland and I don’t have access to oatmeal 24/7? [The team worked from the offices of Highland Capital Partners this summer as part of their startup accelerator program.]
I’m worried about living, and then I’m worried about the business, and then I’m worried about technology and how this is going to go.
BRESEMAN: Are you still glad you’re doing this?
HUANG: Yes. This is definitely one of the best experiences I’ve had. It’s amazing how easy it is to actually do. A lot of our– looking at where we are right now versus where we started the summer, we are so much further than I thought we would get. Opportunities just pop up, and we grab them. It sets us on a path to success, almost.
I went to a talk by the creator of– they’re now called Ink, they were called Filepicker.io– Brett van Zuiden, and he was saying how you can just set yourself on railroad tracks. You make all these goals, and then all you have to do is hit them. And hitting goals is something you’ve been trained to do since you started school: you have a test; all you need to do is pass that test. You have a project; all you need to do is get to the next stage of the project. And along the way, people have opened themselves up to us and have said, “Hey, why don’t you do this? We’ll give you expertise in this field,” and all we have to do is take it. And we take it, and we’re like, “Oh, that was a good decision.” And then that just leads to more good decisions.
I guess starting at Highland was the very first good decision we made, and since then I think we have been setting ourselves up for better decisions. It’s not so much of a “what the hell are we doing”, and more like, “here are the five possible choices; let us pick out the best of these five based on these metrics.”
BRESEMAN: What do you think of starting in an accelerator versus starting on your own?
HUANG: I have one other startup which I tried to do the summer after freshman year, with Charlie Offenbacher. It was essentially a marketplace connecting students with jobs: a less sketchy Craigslist, in a way. Parents needed babysitters; you don’t want to find a babysitter on Craigslist because that’s super sketchy. But you would trust the high schooler who babysat your friend’s baby for your own child. It was more localized, more geared towards parents.
We were working out of Dogpatch Labs, and they gave us an office space, but it wasn’t really an accelerator program. Throughout the three months that I was there, we were iterating on product, but we just never really got anywhere. At the end of the summer we were kind of in the same place that we were at the start of the summer. We were getting a few sales, and our product had developed, but everything else wasn’t really in place. And that was for a few reasons.
One is, the market is kind of hard. Every competitor to Craigslist fails. I don’t even know why we tried. I wouldn’t try that again. Two, I guess– Highland helped us establish a lot of legitimacy around this. I’m not sure if Dogpatch Labs had the same. Because we’re a very technical group and Highland invests in a lot of high-tech companies– so do most accelerators, actually. But working out of Dogpatch didn’t really open us up to that many more contacts than just working out of someone’s basement would have. Especially not in the case of our market, which was parents, for example, would not have heard of Dogpatch; they would not have cared. If they had more experience in nannying services, for example, that would have been a lot more useful.
I guess it really depends on the kind of market that you’re going after, but for us, because we’re students, and we’re young, and we’re building a platform that other people have to rely on, having this sense of legitimacy and have other people think that we’re not just playing around, that this is a real project and we’re actually going to deliver, and doing it from Highland I think has opened up some doors for us.
BRESEMAN: You said before that it felt easy to start Technical Machine. Why do you think it was easy?
HUANG: It was me and Jon and Tim, and we have been working together since freshman year. I’m comfortable with Jon or Tim picking up tasks of mine. I’m comfortable with them taking my tasks, and I think they feel the same with me. I think there’s a lot of mutual trust between us because we’ve been working together for so long on so many different projects.
Some of them have gone really shittily– which is one of the best parts, that things will go shittily, and if I’m at a loss as to what to do, one of them will know what to do. So that was the easy part, in terms of cofounders, and then we brought on some other Oliners. Eric ended up helping us a lot on things; so did you [ed. note: directly addressing the interviewer], on things that we lacked. Things that we didn’t have the bandwidth for. That allowed us to focus on things that we were better at, instead of having to constantly think about, we need press, we need to get all of this coverage. How do we do it?
That was one of our main things that we were not going to do unless you had joined the team, for example. We wouldn’t have had the bandwidth to cover it. Bringing on these people, in these strategic places that you need help in, makes life so much easier.
Hiring is like the best thing ever. I didn’t realize that before I started, but employees are awesome!
BRESEMAN: You’re going back for another semester at Olin– are you going to leverage the Olin community while you’re there?
HUANG: Definitely at least for testing, and I think that we should also be able to leverage the Olin community for some development tasks. There are plenty of good engineers at Olin. Plenty of them are hacking away at stuff anyway, and they’re very much a demographic that we want to hit with our market: tinkerers, hackers, whatever. And I also think that they would be good people to bring on for at least some projects– we need to ship out a ton of peripherals. We want to have a great community in place when Tessel first launches, and I think Olin is a great place to get that started.
BRESEMAN: How does it feel to be running a company and going back to Olin at the same time?
HUANG: Really strange. School feels very familiar; I know exactly what I’m going to be doing in school, and it’s not dangerous, and I get kind of a warm feeling because I’ve been there the last four years.
On the business side, I’m always angsting about Technical Machine. It’s the combination of not a lot of anxiety, because we have a lot of safety in this– just because it’s Olin, it’s like coming home, really. I know that I can be safe, that I can hide, at Olin, from people that I don’t want to see on the internet, for example.
I’m taking like 12 credits. One of them is E! Capstone, which will literally be working on this business; one of them is Chinese, because I’ll be going to China for manufacturing, and the other one is Sanjoy’s Bayesian class, which I heard was good.
I’m not worried about the workload because I got all my hard work out of the way. Hopefully I can just cruise and have a free place- well, a very expensive place to stay while I work on the business.
My mom told me that I had to get my degree. I told her, “Oh, look, we’re trying to raise a round, things are going well–” and she’s like, “I hope you stay in school!” No, mom, God! You don’t understand!
BRESEMAN: Have you seriously considered dropping out?
HUANG: I don’t really care about the degree. I think that I’ve gotten a lot of Olin these past four years, to the point where if I was the weakest link here– if everyone else would have been working working in Boston, in an office together, I would have dropped out. But since Tim also had a semester to finish and he’s not dropping out, and I think it’s important for the team to stay together in this sense, because Tim’s at Olin, I’m going to be joining him at Olin.
BRESEMAN: If Technical Machine were to fail, do you think you would start another company?
HUANG: I don’t know what would happen if Technical Machine failed. At one point, I get really anxious thinking about it because I’ve spent a lot of my time on this, my name is kind of tied to it. My reputation as a developer is almost tied to it because we’re working on a platform for developers.
If it were to fail, I think I would have trouble separating myself out of the product that failed. I really wouldn’t know the next steps that I would take. Eventually, I do think that I would try to start another company. I would try to iterate some more in this space because I do think that it’s incredibly interesting, and the space is growing.
No matter what, the knowledge that I have gained from working at Technical Machine is more than I would have gained anywhere else at this point in my life. Overall, it’s a win-win scenario for me. It doesn’t matter if it fails or if it succeeds. Obviously, it’s way better if it succeeds, but if even if it failed, I would have gained knowledge and made the connections I needed to in order to set myself up for success in the future.