Amongst the capable and talented workers of RunBuggy, we are so proud to have many employees who are considered to be experts in their fields. We met with David Erickson, Senior Data Scientist, to discuss the future of Data Science and his role on RunBuggy’s robust Data Science team.
Why don’t you go ahead and introduce yourself and talk to us about your role at RunBuggy?
My name is David Erickson. I’ve been a Data Scientist for about five years now. I have a PhD in Biophysics from UC San Diego; I also have a degree in Math. I’ve always liked studying complicated problems, which is what drove me to get into Data Science. I did consulting for a few years, some big tech companies, and stuff like that. Then I jumped over to RunBuggy to be a Senior Data Scientist.
Can you talk a little bit more about where you worked before RunBuggy?
Just before RunBuggy I was at a consulting company in San Francisco called Qbiz. They do only data consulting. So I did projects with Levi’s, I did projects with Facebook and Cash App. I had a great kind of all-around introduction to the industry. I started with RunBuggy on Valentine’s Day (2022) so I’ve been with RunBuggy for almost exactly seven months I guess.
Could you tell me a little about what the Data Science team as a whole does at RunBuggy?
We have a pretty large Data Science team. I’m sure as you know, RunBuggy is a technology company. We’re technology focused and we want to use our data to the biggest extent possible, so we have quite a few projects going on right now. As always, we have a lot of infrastructure projects. Grabbing new data sets and integrating with data sets we already have access to either within our own software or third-party vendors that we work with. Yeah, we have a wide variety of data science projects using that data. A lot of stuff is around helping RunBuggy and the various teams at RunBuggy make data-driven decisions. So measuring the things they want to know about, providing KPIs, following up and doing analysis’ of, how to improve in certain ways. A lot of that boils down to helping our customers. So we, the Data Science team, put a lot of effort into thinking about the haulers that we work with and the shippers that we work with, and how we can serve them better. So often things like “How can we move an order faster?” and “How can we make sure that our orders fill trucks as much as possible, which decreases costs, makes more money for the haulers, saves on wasted energy and carbon emissions and everything like that.” A lot of projects looking around and seeing how we can help everybody.
Would you mind expanding on what your day-to-day responsibilities are in your role?
The Data team, we’ve all kind of got our specialties. We work on things a lot together but we all specialize a little bit. I do a lot of work on growth and emails and generally helping to make decisions with data. Growth usually means not only how can we help RunBuggy grow, how can RunBuggy grow more, but often what that really means is how we help our shippers grow more with us, and how do we help our transporters grow more with us. I put a whole lot of effort into that part. Helping to make orders move more smoothly, and make sure everybody is happy so that we can all grow together.
Could you tell me about how Data Science can improve transportation logistics?
There are a lot of ways, obviously. There are the traditional ways which usually just get called Logistics or something like that. That is a more old-school industrial model. Things like hub and spoke systems or traditional Transportation Engineering. We actually have a Transportation Engineering PhD on our team. We go beyond that and apply state-of-the-art algorithms, and Machine Learning, and Data Science principles to help move everything a bit quicker. We try to include all of the relevant information that we can to help everybody grow using Data Science. I think traditionally moving vehicles is a tough industry. You’d get an order in, and then you just pick up the phone and start calling people. Maybe you have a bit of intuition or knowledge about a particular hauler likes these certain types of routes under these conditions. What we are able to do in Data Science is kind of automate that whole process to some extent. Obviously, we can never replace our amazing Ops team and Customer Service team. We are able to help them a bit and we’re able to kind of identify as orders come which haulers are most likely to want to move those orders, and how we can move them most efficiently. We use not only these kinds of traditional methods, but we also consider that we use state-of-the-art Machine Learning, so we look at all the data we have. We think from a transporter’s point of view, “What do we know about this particular transporter? What kind of moves do they like to do? Do they like to do oversized vehicles or non-operational vehicles? Do they just like doing standard, regular-sized, operational things? Do they like delivering to businesses, or do they prefer delivering to homes?” That kind of thing. We integrate all of this into a model and predict how good of a match each order is with our various haulers. Then we can proactively reach out and say, “Hey, we noticed that you have extra capacity right now and we have some orders on our platform that are perfect for you.” We can kind of do that instantaneously, so that’s been very successful here.
What does RunBuggy’s Data Science team to do help eliminate empty miles?
We do a lot there. It is a really tough problem, as everybody knows. We’ve got a Transportation Engineer PhD on the team, who was actually a professor in the past- Saber Abdoli. Saber has been doing a lot of work on those algorithms. We have an idea of what the capacity of various trucks are. Then we do a lot of analysis on road systems and transportation networks as we have orders come up to say, “What is the likelihood that these two orders can be put on the same truck? Are they close to each other? Are they close to each other not only geographically but on a road system? And are they, does it make sense every step of the way for a truck driver to go ten miles out of their way to pick up this new route, whatever it is?” That is kind of the foundation of our work. We continue to add layers on top of that about various preferences transporters have. Maybe they like to pick up orders in certain locations, but they don’t like delivering to certain areas, whether that is particular cities or certain very rural areas, or whatever that is. We can kind of monitor all of those preferences, and put together a pretty good model of which transporters are interested in various orders. Once we do that we can start proposing various groups of orders to the transporters. We know that they have a very hard job. At the end of the day, we aren’t going to be able to replace the hard work that they do in planning their own logistics. What our goal is right now is to provide them with the best information and the easiest way possible to help them group the orders as much as they can. If we know someone has a nine-car truck, we try to make sure they can get nine cars on it.
Can you speak on algorithms and automation’s impact on the environment?
Algorithms and automation are pretty central to the way that the RunBuggy platform works. We’ve integrated these various ideas. We explore different ideas algorithmically, we test them all before we put them into our production system on our platform. We have quite a few that are working very well right now. An example of an algorithm is that we kind of monitor the state of the platform, and how many open orders we have, and where they are, and things like that. Most of the time our platform works very well. We don’t have to do too much intervention in terms of getting orders moved more quickly or alerting people to their existence. But if we suddenly have many more orders than we did yesterday, then often transporters aren’t checking every second, they don’t realize that we have so many orders in our system. What we’ve been doing recently is proactively reaching out. We say, “Hey! Oh! We have a lot of orders.” The algorithm recognizes it, it runs, and it puts together various groups, and matches those with different transporters. Then we proactively reach out to them and say, “Hey! We’ve got a few full trucks’ worth of orders for you. Would you like to take them?”
Would you mind telling me about load bundling? What impact does it have on transportation costs and carbon footprints?
Our goal is to have every truck completely full for as many miles as it can be. That is kind of the idea of load bundling, or load grouping, order grouping. You can think about it in a very simple way. Let’s say you have a truck that is hauling one car, and it hauls it across the country. Well, then the truck driver still has to get home. They also have this kind of backhaul it is often called. Ideally, they would want to carry an order back across to the other side of the country with them so that they aren’t driving with an empty truck. If they do that, they are effectively getting two orders done. Double the pay, but driving the same amount of distance effectively. On the opposite end, if we aren’t able to group an order or find a backhaul for a hauler, then they have this situation where they drove all the way across the country, they have to drive all the way back empty. You can see immediately that they are only making half the money, they are still buying the same amount of gas, and they are still producing the same amount of CO2 into the atmosphere. Everything we do there helps not only the environment and the entire logistics sector, but especially the transporters who make more money and the shippers who get things moved more quickly and easily.
Can you say what allows RunBuggy to reduce the cost and time to ship a vehicle compared to our competitors?
I think our Data Science team is a part of it with all this order grouping. Any of this load bundling helps tremendously. A huge part of it is that we don’t place any constraints on any of our transporters or anything. We are an open platform, we accept all orders. Not only this grouping, of course, which everybody tries to do. I believe we’re a bit better at it because we have a great Data Science team and we have orders from every state in the country. We work all over the place, so we have a giant network of shippers and transporters. That alone helps a lot. Our software, our platform makes it easy to group orders. As we’ve grown, we can see that the more orders we get in the system the easier it is to begin to match orders and make things as optimal as possible. A huge part of it for the Data Science team is that we’re constantly monitoring things. Of course, we’re constantly learning and improving everything. Whether that is algorithms, or what our customers want, whatever their preference is.
What other factors make RunBuggy better than competitors in the same sphere?
RunBuggy is pretty unique, we are the only platform. We work with an open set of transporters and shippers which gives us more flexibility than everybody else. We certainly invest more in technology and data-driven decision-making than everybody else. Other companies may have, a few other companies have Data Science teams but nobody has a Data Science team like RunBuggy. I think we’re up to eight people now. Three of them are PhDs. We’ve got people with decades of experience in computational modeling of complex scientific processes. A former professor of Transportation Engineering. That is a huge part of it. I think really a lot of it is how we focus on our customers. I think for most of our competitors, they are kind of a broker. They don’t care that much. They are a message board. They are something else. But for us, it is all about our customers. We only grow if our customers grow. We put a lot of effort into learning what our customers like, what they don’t like, what keeps them on the platform, what drives them away from the platform, and how to make them the most money.
Having a lot of experts on our team poses RunBuggy in a better position than our competitors. As one of these experts in your field, what do you think the next five years look like in Data Science and Automotive Logistics?
It is going to be a really exciting time in Automotive Logistics. As I’m sure you’re aware, many people are aware, the automotive industry is undergoing huge changes right now. There is massive growth of the electric vehicle sector. There is a large push away from the dealership model and towards the home delivery model for electric vehicles. This means much more difficult logistics. It is a lot easier to ship a load of cars to a dealership than it is to individually ship each car to the final owner of that car, wherever they live in the United States. That presents a lot of opportunity for RunBuggy, especially in the Data Science realm. All of the logistics, everything about orders, everything about efficiency. These are core Data Science problems that we will be able to help optimize and make everyone as happy as possible.