Racing at StashAway
October was such an exciting month for us at StashAway! We held our first in-person offsite since 2019 at our KL office! It was an action-packed month with a combination of work and fun, spending time together, creating bonds with people we’d only seen on screens before, and living and experiencing the company’s culture to its fullest.
As part of this offsite, with the help of AWS, we also organized a special event for our data and engineering teams, the AWS DeepRacer. The goal of the DeepRacer is simple: train a machine-learning (ML) model, upload it to the race car, and achieve the fastest lap during the physical time-trial event on race day!
We split our engineering team into groups, and team “Wealthy Wheels”, “Precision Pilots”, “Data Drifters” and others got to work. First things first — let me take a selfie! Some teams got creative and leveraged AI before the event had even started:
In the subsequent weeks the teams got to use the provisioned AWS account with some allocated training hours to train their models on a virtual track. To ensure everyone can have some fun, the AWS interface makes this very accessible for beginners, while still allowing a fair amount of fine-tuning for those who want to go deeper (pun intended…)
Time for race day!
The stakes are high, and everybody is performing some final adjustments to their models in anticipation of the upcoming race. The physical environment is different from the virtual one, as you have to deal with reflections on the track, little crinkles on the carpet, and humans who are running behind the cars that go off track to put them back on their path. Which means that a fast lap on the virtual training ground does not guarantee success on race day!
After a total of 105 laps raced, we have found our winning team! With a fastest lap time of 8.6 seconds team “Wealth Wheels” won the event and secured 1st place! Fun fact: the slowest lap took a whole minute to complete — definitely a lot of work for the race marshal who had to run after the car and help put it back on track 😬
The DeepRacer event was not only a learning opportunity, it was also a whole lot of fun! The event provided all of our engineers with a unique opportunity to explore the fascinating world of Machine Learning (ML) and Reinforcement Learning (RL), to build and train models to race their cars, and experience first-hand the challenges of getting those “to production” (aka try to stay on track and not crash into the barriers…)
Thank you Amazon Web Services (AWS) for being our partner in this exciting journey. We look forward to many more exciting collaborations in the future!
