What a ride! The DeepRacer league is ending but was an amazing journey
This post describes my experience racing on DeepRacer since 2020 and how this technology tool empowered community building both locally and globally
Published Oct 31, 2024
AWS DeepRacer is a service that lets developers create Machine Learning models for autonomous racing. The models can then be loaded into physical cars for racing on a track or run in a simulator for virtual racing. Inside the AWS Service, developers can compete against each other for prizes and fun. That virtual competition ends today (October 30th, 2024), and I'm writing this post 3 hours before it closes.
This is a bittersweet moment for me and many other racers who made DeepRacer their primary hobby. On one hand, it is a sad time because we will no longer be able to participate in it. On the other hand, I'm grateful that DeepRacer helped me learn ML, connect with individuals worldwide, and create a local community that will continue to grow while racing.
Personal Growth
DeepRacer announced a F1 challenge in 2020, during the pandemic, and that immediately caught my attention. I was an engineering manager with a lot of experience in development and infrastructure roles but no exposure to ML. DeepRacer helped me grow that skill, and with time, I became proficient: I'm now the CTO of an AI startup, teach at a local university, and collaborate as a Certification SME on the AWS AI Practitioner and MLE Associate exams. I also developed some non-technical skills: I became more methodical; every day for four years, I had to try a new model, see what worked and what didn't, and think about what I could improve. Sometimes, this was hard, frustrating, and very challenging, but the rewards have been amazing.
Global Racing
In 2020, when I started experimenting with DeepRacer, I made it to the finals in Las Vegas at AWS re:Invent. However, since this was during the pandemic, the event was held virtually.
In 2021 and 2022, I continued to compete, but the competition became harder, and I couldn't make it to the finals.
In 2023, the competition rules changed; finalists were now split in different geographies, allowing 8 finalists from each global region. This generated a lot of controversy because some regions had top racers (like Asia, the US, and the EU), and the 8 slots generated a lot of competition. It seemed unfair that someone from a region like mine, LATAM, could become a finalist without having a great race. However, I think the point AWS was trying to make was right, as this helped engage and expand to regions where DeepRacer was not yet popular. Nevertheless, participating in this final was an enriching experience for me. It helped me connect with individuals from all around the globe whom I had previously been able to meet just virtually.
You can read more about my experience as a finalist in 2023 here.
Local Racing
When I qualified as a finalist in 2023, I took it as an opportunity to grow our local DeepRacer community while learning the physical aspects of racing, something I hadn’t experienced before I collaborated with Pablo Inchausti; he and I are both AWS Community Builders and saw the chance to start doing races at UADE, a local university, and build a local DeepRacer community for both the university students and the local AWS User Group. We did a lot of racing in 2023, continue doing it in 2024, and we plan to keep doing it in 2025.
Conclusion
DeepRacer competition is ending, but the technology will become open source, and universities and companies will be able to continue using it and learning from it. I'm grateful this tool helped me grow technically and has helped me teach ML to students and community members.