Pre-Graduation Quarter Syndrome occurs in students either entering or in the last quarter of their degree program.
After covering the topical breadth of their major coursework, the student may have honed in what topics do and do not excite them. This allows for their coursework to be exciting and meaningful in the development of their career. Filler classes seem to be fewer and farther between, as the student has progressed through major requirements and on to the elective and senior coursework.
Despite the hectic nature of the job search and the uncertainty that the future holds, this quarter is lining up to be pretty awesome. I had initially planned on transitioning my summer internship’s work on to my senior capstone project, but when I saw the opportunity to work on a reinforcement learning project, I couldn’t pass it up.
To give a little background on why this project is so exciting to me we have to go back to my senior year of my first undergraduate degree, in 2019. I had already decided to minor in computer science at that point, but I had not really decided what to focus on within the major. That’s when I saw this awesome video from OpenAI. OpenAI, Elon’s tech company devoted to the development of AGI (artificial general intelligence) for societal good, created a playground in which agents were to learn how to play hide and seek. These agents, hiders and seekers who’s roles I’m sure you can deduce, learned the strategies of the game without human interference or guidance. The hiders learned how to move blocks in order to create enclosures that the hiders couldn’t walk or see into. Subsequently, the seekers learned how to use ramps to invade the buildings that the hiders had created. The seekers even figured out a bug in the physical laws of the game eventually and exploited it to attain victory.
These agents were learning and, what appears to be, thinking. Back when I saw the video for the first time it was absolutely mind blowing to me, and to be honest, it still is. So when I saw the chance to create something similar, I had to jump on the chance. My group is making a clone of the old Atari game, Breakout. We plan to create an agent that can play the game and learn the tactics of the game. Eventually, we plan to pit the agent against a human to compete, but I’d also like to stretch the game a bit to see how different generations of our agent compare to each other. I want to see how the strategy of each agent evolves if we put two agents on the same board and with their own ball and see who can break the most bricks.
I’ve always planned to eventually create such a project with soccer, my childhood sport and hobby. I’d love to see how agents learn to play the game and what strategies develop over time. This project is propelling me in the right direction for creating my own simulation of the sport and eventually reaching my goal of working in multi-agent reinforcement learning.
Outside of my senior capstone project, I’m taking the cloud development class. This course is pretty awesome, as it has covered RESTful API’s, the bread and butter of backend software engineering. I’m getting great experience with reading documentation and incorporating pre-built software into my own applications. It’s something highly relevant and eye catching when placed on a resume.
Lastly, I’m still continuing work from my summer internship throughout the quarter! I’ll be assisting with the wildfire classification and detection models, and how to cleverly use mobile robots to assist with these tasks. Not every undergraduate has access to real world data, and gets to train machine learning models on it
I feel extremely blessed for these opportunities and I’m glad my time gets to be devoted to them rather than the facets of computer science that don’t interest me.