• Languages/Frameworks/Software: Python, C, Scapy, subprocess, ssh, Raspberry Pi, Grafana, Z Wave, InfluxDB

• Developed a sophisticated energy management system for Swarthmore College's computer labs, leveraging ZEN15 smart switches and Samsung SmartThings Hub to reduce idle power consumption from 20-50W to 5W per computer.

• Engineered a local agent using Python that dynamically applies sleep policies based on user activity, CPU usage, and time of day, significantly enhancing energy efficiency and system autonomy.

• Implemented a robust remote access solution, utilizing Wake on Unicast Traffic and a Raspberry Pi-based ARP relay to enable seamless remote wake-ups, ensuring continuous availability without physical presence.

• Designed a comprehensive monitoring infrastructure with InfluxDB for real-time power consumption tracking, and developed a public-facing website to visualize energy savings and advocate for ongoing environmental sustainability initiatives.

full project description

• Languages/Frameworks: Spotify API, NLTK, scikit-learn, matplotlib, jQuery, d3.js, Flask, Python

• The project uses d3.js and NLTK to create an interactive visualization of Spotify playlists, representing songs as animated circles based on various criteria like genre, popularity, and energy.

• The backend script employs the Spotify API and Flask to query playlists, process track information, and handle OAuth2 authentication, storing the results in JSON files for further analysis.

• Lyrics from each song are obtained using the LyricsGenius API, preprocessed, clustered using K-Means, and visualized with TF-IDF vectorization and PCA to reveal patterns and themes within the lyrics.

• The frontend interface uses D3.js to visualize song attributes in two SVG components: a main SVG for genre, popularity, and energy, and a smaller SVG for detailed exploration, with interactive controls and dynamic updates enhancing user engagement.

full project description

• Languages/Frameworks: Flask, Numpy, NLTK, scikit-learn, GloVe, UMAP

• Animated interface of creative text exploration generating a continuous sequence of sentences, linking each to its most semantically similar line in a 2000+ line corpus from 12 female mystics.

• The backend leverages the Flask framework and multiprocessing to generate prayer sentences in real time using a combination of natural language processing (NLP) techniques and GloVe word embeddings, and employs a JSON API to asynchronously serve these sentences to the front end.

• The backend also utilizes locality-sensitive hashing (LSH) and UMAP dimensionality reduction to process and analyze prayer sentences. It embeds sentences into a semantic space, identifying relationships and generating a morphing sequence of sentences, which are then visualized on the front end.

• The front end employs custom JavaScript functions to create an engaging user interface. The script dynamically positions sentences in a zigzag pattern, adjusts font sizes based on scroll position, and adds interactive flower elements that correspond to unique words. It also visualizes prayer data as points in a 2D space, with features like scaling, connecting lines between points, and an interactive slider for enhanced user interaction.

full project description

an interactive metaphor for that one famous Walter Benjamin quote

full project description

an ode to two things I miss from my school

$10,000 scholarship awarded to Computer Science students with disabilities

• under mentorship of Danae Mexata and James Landay from Stanford HCI, I conducted critical analysis on the politics of digital visibility and vulnerability within social justice movements, analyzing case studies including Black-owned business labeling on Google Maps, ICE's identification of undocumented immigrants through social media, OnlyFans' ban on sex work, and YouTube's unintended demonetization of LGBTQ+ content creators, contributing to the understanding of marginalized communities' experiences online.

• Developed a comprehensive framework to assess the tension between visibility and vulnerability for marginalized groups on social media platforms, categorizing case studies based on their impact on marginalized users' visibility and platform designers' intentions, providing valuable insights and recommendations for platforms to address the potential harms faced by marginalized users.

• Authored an extended abstract titled "Visibility & Vulnerability: Understanding Marginalized People's Experiences Online," highlighting the negative impacts of social media on marginalized communities and offering recommendations for platforms to better serve marginalized users. Emphasized the importance of involving marginalized perspectives, providing additional resources and support, and aligning platform values with user well-being, contributing to discussions on mitigating disparities online.

• Languages and Frameworks: Selenium, Beautiful Soup, Pandas, geocoder, Flask, Folium

• Designed and implemented a comprehensive search engine and geovisualization tool compiling datasets of healthcare resources nationwide for sexual assault victims to 1) identify the nearest hospitals equipped for Sexual Assault Forensic Examinations (SAFE) 2) locate sexual assault professionals including IAFN-Certified Nurses, sexual abuse therapists, social workers, and professional counselors by scraping together a custom database consolidating sexual assault professionals by area from four public domains.

• Utilized web scraping and automated browser interaction to extract data from dynamic web pages, while managing HTTP requests and processing user inputs to enable real-time, location-based data processing and visualization.