As part of my MEng in Computer Science at MIT, I completed my thesis project in collaboration with the Space Enabled group at the MIT Media Lab. The project investigates how Instagram’s content recommendation systems mediate visual representations of identity. Analytical and deep-learning based methodologies are leveraged alongside a historical accounting of photography, media representation, and algorithmic mediation to provide a race-critical audit of dominant archetypes and controlling images within gay-coded spaces on Instagram. The paper concludes with a discussion on how we might create empowering visual representations and more equitable social media platforms.