When you try to fit everything that a car dealership does into a mobile app, you better believe there’s some extensive tech and data that goes into it. For Blinker, the effort to do just that is built on cross-team collaboration and enthusiasm for putting the car-buying process into the hands of regular people.
We spoke with three senior members about how they use proprietary algorithms, machine learning and computer vision to gain valuable insights about the industry and challenge the car-buying status quo.
WHERE THEY DO IT: Denver
WHAT THEY DO: Blinker is a free mobile app that’s reimagining the process of selling and buying cars. With image-recognition technology, Blinker puts users in the driver’s seat, helping them buy, sell and refinance their cars with the snap of a photo.
THE IDEAL CANDIDATE: An empathetic self-starter who’s a good communicator and problem-solver. Team members at Blinker want to roll up their sleeves and get things done each and every day.
SIX DEGREES OF JOHN ELWAY: Blinker’s founder and CEO, Rod Buscher, also founded the Denver-based John Elway Auto Dealerships in 1989.
LEADING BY EXAMPLE: VP of Engineering Jimmy Bartolotta joined in 2015 to accelerate efforts to answer one central question: How can tech totally change how cars are bought and sold? Learn more.
Solomon White, Chief Architect
As chief architect, Solomon is the technical compass for the engineering team, steering it in the right direction by recommending technologies and tools and ensuring the maintenance of software systems.
BEYOND WORK: Solomon likes boarding, whether it’s to work on his longboard during the summer or down the slopes on his snowboard during the winter.
Can you tell us how data analytics and machine learning play into what Blinker is doing?
We analyze event data to see how customers are using the app and find areas that seem to be the most frustrating. That way, we can focus on improving UX where it will have the maximum impact. Additionally, we are applying analytics and machine-learning techniques to vehicle sales data to improve our understanding of the market value of vehicles, loan performance data to build better risk assessment models and conversion data to optimize our marketing spend.
There are many novel ideas to work on and problems to tackle that keep me enthusiastic about coming to work every day.”
What’s in your tech stack, and why did you choose it?
We try to be pragmatic about our technology choices. We chose Ruby on Rails as our main API framework because it strikes a good balance between developer productivity and application performance. As we continue to grow, some performance-critical portions of the API will inevitably become microservices in Elixir or Go.
For our computer vision and machine learning projects, we use Python, Tensorflow and Keras, due to the wide variety of network architecture implementations. Our mobile applications are doing enough edge processing — for example, machine learning predictions — that it makes sense to maintain them as separate native applications, so they are in Swift and Kotlin. We use Docker and the HashiCorp suite of products to ensure that we have reliable infrastructure for all our server-side tech, and AWS and Google Cloud to enable infinite scale.
What are you currently most excited about at Blinker?
I enjoy learning new technology and delivering business value, so the opportunity to practice machine learning, and especially deep learning, is very exciting to me. We’ve built some interesting technology, like our vehicle framing model that can determine how well an image is centered or cropped around a vehicle for an ideal listing photo, and our salability score that can predict the likelihood of a vehicle selling. There are many novel ideas to work on and problems to tackle that keep me enthusiastic about coming to work every day.
Scott Meschke, Director of Engineering
As director of engineering, Scott is responsible for his team’s overall performance and productivity, ensuring his team is effectively cross-collaborating and integrating with other teams at Blinker, and assisting his VP and chief architect.
BEYOND WORK: Scott regularly writes programming-related articles on his personal blog, which he says has even helped his verbal communication skills along the way.
How do your past professional experiences compare to what you’re doing now at Blinker?
I started my career in Salt Lake City. I have mostly worked for fintech companies, spending most of my time developing Android apps and sometimes doing back-end development in Java, Ruby, Elixir and Go. I have also been a team and project lead before, but this is my first time operating as senior leadership within an engineering team. It’s been great to step into a new role and have support and guidance from leaders across the company.
Tell us about a project the Blinker team is working on that excites you. How are you taking it on?
I am a huge believer in building tools that empower knowledgeable people inside the business to operate and influence business outcomes without expensive and highly contested engineering time. We are currently building out a syndication service that will allow non-engineers to adjust outgoing syndication schedules, number of cars, market locations and more.
While those changes used to take at least 10 to 15 hours of engineering work a week in the past, the new syndication service will allow our marketing team to not only own those decisions, but to evolve their syndication strategies and specific details without engineering support.
Blinker puts a lot of effort into hiring the right people, and that pays dividends.”
What’s one thing about Blinker that you really love?
I really enjoy spending time with my coworkers. There are often events and gatherings happening outside of work, which are always fun to attend. There’s a sense of friendship and camaraderie among departments and teams but also across traditional org-chart boundaries. Blinker puts a lot of effort into hiring the right people, and that pays dividends in the day-to-day operations of the company.
Chris Leone, Manager of Data Analytics
As the man behind the data, Chris is responsible for ensuring that Blinker is able to measure and track every aspect of their business and turn that data into actionable insights shared through various reports, dashboards and in-depth analyses.
BEYOND WORK: Though it’s nearly impossible, Chris tries to take advantage of all the outdoor adventures Colorado has to offer through biking, hiking (with his fiance and three dogs), fly fishing, snowboarding and rock climbing.
We’ve heard you’re kind of a jack-of-all-trades, handling everything from business initiatives to analyzing data. Does that speak to the kinds of collaboration that takes place between teams?
What makes my job so enjoyable is the fact I get to be involved in so many different aspects of the business. It can be challenging with context shifting and keeping track of everything, but the sole reason I’m able to cover such a wide spectrum of work is the collaboration. With this sort of cross-team support, you not only gain efficiencies by sharing the workload and knowledge from different teams, but it’s downright fun to unite individuals from different parts of an organization around a shared goal.
[The] sole reason I’m able to cover such a wide spectrum of work is the collaboration. "
What’s a technology or data set that you’re working with at Blinker that you’re particularly excited about?
Getting to use data that exists nowhere else and that was designed and created just for Blinker is truly unique, and the use cases are limitless. The fun part about data is that from the moment you start working with it, the questions and opportunities only grow.
For example, with the seller metrics data that our computer vision and machine learning team created, the primary goal was to determine if we can predict the sellability of a vehicle. We have combined that data with other measures to create a marketing score to aid in marketing efforts. We’ve also built our own attribution methodology that allows us to know the source someone originated from, which gives us data to compare performance from the very top or very bottom of a user journey by source. If you haven’t noticed, the general theme I’m most excited about is continuing the evolution of internal data solutions.
If you look at what your team has achieved so far, what are you proudest of?
When I started back in May 2017, there was no data team and no structured databases, but there was still a very strong appetite for data. This meant engineers were running queries against the production database in order to view any sort of data or reporting. I started around the same time as our former data engineer, so we both had to hit the ground running to work on tool evaluations and then get them stood up to start surfacing data.
We signed with Chartio, and, from that point, we’ve created more than 120 dashboards and automated reports, as well as had the ability to surface information and dig deep anywhere necessary. The work obviously doesn’t stop there, but I’m extremely proud of where we got in a short amount of time and even more excited about where we’re headed.