AWS approach is unconventional compared to other learning formats. Ambiguity and non-prescriptive guidance allow teams the flexibility to think creatively as they navigate a wide array of technical challenges. There is no one right answer; teams pave their own path based on resources we provide them with in live AWS accounts. Assisted by AWS and Firemind.

The setting for every GameDay is a famous (fictitious) narrative. Participants are all new hires at Unicorn.Rentals and this is their first day on the job. They’ve been left to take over a number of complex problems affecting the business.

The event began with teams taking their positions and each coming up with their team names. Below you’ll find the selected names for the day:

The Quants, Grip it by the Horn, You need corn… now!, Phat Sine.wav, Tao AI, Sparkles, Goofy_bhabha, Looper-Da-Loop, SageBreakers and Phat Unicorn Wave.

Once the teams had set their names for the event scoreboard, it was straight into the action by getting the right access to the AWS environments. All contestants would be using AWS services to complete the following tasks throughout the day:

• Step 1 – First things first, it’s time to create a unicorn Forecast. The teams had to create a dataset of unicorn demand data as well as define the algorithm and train the model.
• Step 2 – Pipeline to Profit Town! All teams would need to use Amazon Sagemaker Studio and Lambda functions to train and deploy model endpoints. They’d then need to make smarter improvements to the entire workflow, scoring more points when SageMaker Inference Model Endpoints were added.
• Step 3 – Things are heating up now! The final stage of the competition was to build a chatbot to help take Unicorn orders. This final task would require concentrated efforts in utilising Amazon Lex V2 Console as well as Amazon Kendra.

During the first hour, all teams were slowly making their way through the reading materials and acquainting themselves with the AWS services and environments. All teams stayed around 0 points until The Quants, You need Corn.. now! and Looper-Da-Loop jumped into the lead with 15,000 points.

As the competition progressed, Firemind and AWS Solution Architects were at hand to assist with S3 locations, AWS service use and more. It wasn’t until 2 hours in that the pace began to quicken, and Goofy_bhabha jumped into first place (with 50,000 points), defeating the first task and moving onto the second.

Fast forward to 30 minutes before the end and all teams are flying to score as many points for each task completion as possible. That’s when it happened, The Quants managed to score a hefty 25,000 points right before the timer expired!

The final scoreboard revealed the podium finishers for the 2022 ML GameDay as the following:

• 1st place – Looper-Da-Loop
• 2nd place – The Quants
• 3rd place – Sparkles

The team achieving the best name award was The SageBreakers.

Congratulations to all the teams who took part and battled their way through an exciting Machine Learning experience, using AWS services.