The game was designed to implement the 'Human in the loop' machine learning model. We approached this problem by designing a game based on natural disaster management. The goal of the game was to have the player work together with an algorithm called Debra to manage the running of 3 natural disaster management companies and their trucks. The player would create a solution (assign truck routes), Debra would analyze the solution and offer suggestions. Players can accept the suggestions or create a new solution. The process was iterative to find the most efficient solution which entailed lower intersection between different company routes, lower time to collect and dispose of debris and maximize profit.

The team consisted of 6 members. I served as both producer and programmer during my summer internship.

• As the producer on the project, I was in charge of setting weekly and monthly goals. I also set an end of summer goal after many lengthy meetings with the team. The project was an ongoing undertaking and we were asked to meet certain criteria before the end of the summer. I was in charge of heading meetings during the week, facilitate communication between different team members, update design documents, and checking deadlines and make any adjustments if required.

• As a programmer, I was tasked with adding functionality to the user interface. This included designing the UI, adding functionality to buttons to select certain areas of the map, connect the backend algorithm to the main game, etc. I also implemented the tab system which allowed the player to work on multiple instances of the map each with their own solution along with a system to record at least 5 iterations. The user's play data such as areas selected, profit made for each company, route timings, number of intersections between different companies and their routes. This was done on the client-side using JSON.

By the end of my internship, we had created a well-documented design and implemented the basic functionality. Our work was picked up by the next team, and the game was finished and used for research in April 2019.