Max Goldberg joined Kargo in Spring 2021 as the company’s first Computer Vision Engineer. We had the opportunity to sit down with Max and had a great conversation about his role and what makes working at Kargo so special.
I joined Kargo because in my last role I was doing primarily project-based work, working on small, individual components of a system. Prior to that, at Garmin, I had built some really cool computer vision systems end-to-end. So, coming to Kargo, I was looking to get back to that and carry something through end-to-end.
Day-to-day is highly variable. Some days I’m chasing down where valuable and new sources of information for our data sets are going to be. Other days involve setting up the infrastructure for the training and evaluation of deep learning models. More recently, my days have involved building some infrastructure to be able to essentially emulate the operation of the system in a prototype environment, and then designing algorithms around that.
On the software team, everyone is an expert in their role. Everyone is very experienced and has good judgement on what to build. I would say we are – I don’t know if this is too cliche to say – I think we are a very scrappy team. In terms of the problems we are trying to solve and what we have been able to accomplish so far, we are punching above our weight.
This may also sound kind of corny, but I feel like I’m able to live up to (or attempt to live up to) my potential as a vision engineer here. In fact, I need to in order to succeed. That’s an opportunity I didn’t have at my previous positions. Working at Kargo has given me the opportunity to grow faster than I was before.
I can’t overstate how important the personal growth aspect is to me. My first job was strictly a research position. Very slow-paced, without much incentive or need to produce a finished product. In that role, after a long period of prototyping, I was able to build full systems on my own, but they weren’t directly productized or someone else ended up taking them over the finish line. Then the other places I’ve worked have been more execution-focused. So working at Kargo is really the culmination of both skillsets I gained over those experiences.
Hmm I’ve learned a lot. Part of the reason I joined was because I saw much of the infrastructure had already been set up to be able to test and simulate the operation of a system. Sometimes, algorithmic breakthroughs can happen merely by virtue of having the right kind of testbed for an algorithm. Basically, having the ability to reproduce what is going on in the production environment can, in some cases, make it obvious what the next move is. If you don’t have that capability, you’re guessing what is going to work. The ability to actually see how the production system is operating, and then to quickly iterate on algorithms, has unlocked a lot for us recently.
There are two facets. I would say the work I’m most proud of, from an applied research perspective, is the recent label tracking stuff we are doing. It combines deep learning and robotic state estimation in an interesting way. And it is unlike anything else I’ve seen out there.
The other thing I’m proud of is that we set up version one of the system in 3-4 weeks. The end-to-end complete wide angle understanding of the scene. So I’m pretty proud of our ability to do that.
From an engineering perspective, I would say that people are very honest in terms of what can we do today, what is going wrong, how can we fix it. That’s one thing I like about the culture, I can say “hey I think we should do this” and it is seriously considered.
I believe success requires that you both know when to take a step back, to think, reflect, prototype and explore solutions, and then when to not let perfection be the enemy of good, and execute on the vision. That is the very unique opportunity we have here. Usually you are doing one thing or the other.
I used to breakdance in high school. That is a real true fact that no one knows about me.