RAWCSAV

Background

Primary

Secondary

Tertiary

Text

Accent

Home About Projects Visual Writings Contact
RawC-On Project ImageLavaShuf LogoMac Screenshot OCR Project ImageBotify Project ImageBotify Grime Project ImageNew Utility Project ImageHinterview Project ImageArcGIS - Paris Project ImageGoGuard Project ImageSocks Toggle Project ImageNYT Crossword Reset Image

Rawc-On analyzes your Spotify data to reveal deep insights about your music preferences. By tracking audio features like danceability, energy, and mood across your favorite tracks, it creates interactive visualizations of your listening patterns. The platform offers detailed analytics for music enthusiasts and casual listeners alike, with features spanning playlist analysis, genre trends, and personalized recommendations across different time periods to help you better understand and explore your musical journey.

Python Icon Python Flask Icon Flask HTML Icon HTML CSS Icon CSS JavaScript Icon JavaScript MySQL Icon MySQL Celery Icon Celery Redis Icon Redis NGINX Icon NGINX
RawC-On Project Image

LavaShuf enhances Spotify's shuffle feature by replacing the pseudorandom algorithm with true randomness. Inspired by Cloudflare's LavaRand, the system captures video of a lava lamp to gather entropy, processes the video frames to detect and track keypoints, and then uses these data points to generate random numbers. These numbers are used to shuffle Spotify playlists, ensuring a genuinely random and fresh listening experience. This methodology leverages a sophisticated pipeline involving a trained YOLOv8 model for detecting regions of interest in the lava lamp video, ORB for tracking keypoints within these regions, and secure hashing with SHA-256 to produce unbiased random indices for song selection.

Python Icon Python YOLO Icon YOLOv8 OpenCV Icon OpenCV PyTorch Icon PyTorch Spotify Icon Spotify
LavaShuf Logo

Mac OCR is an application designed specifically to tackle the academic obstacle of those files not conducive to typical text editing. Integrating with the native macOS screencapture, it recognizes plain text or scientific notation (converted to machine-readable LaTeX) within the image, and appends it to the clipboard for easy copying. The application leverages Google's optical character recognition (OCR) algorithms to accurately interpret text from various image formats. It also supports multiple languages and can handle complex formatting, making it a versatile tool for students, researchers, and professionals who frequently work with scanned documents or images containing textual information.

Python Icon Python Pytesseract Icon Tesseract LaTeX Icon LaTeX macOS Icon MacOS
Mac Screenshot OCR Project Image

Botify was a personal project in finding the best manner of segmenting a Spotify playlist using a combination of genre embeddings, numerical audio features, and lyrical descriptors. These categories are tested using different weights, various dimensionality reduction techniques (PCA, t-SNE, and UMAP) and assorted clustering algorithms (K-means, Agglomerative Clustering, and Spectral Clustering). Tests were evaluated using metrics such as Silhouette Score, Davies-Bouldin Index, and Calinski-Harabasz Index. It also includes fun, thematically coherent art generation for each playlist.

Python Icon Python Pandas Icon Pandas NumPy Icon NumPy Matplotlib Icon Matplotlib Seaborn Icon Seaborn Cloudinary Icon Cloudinary Spotify Web API Icon Spotify OpenAI API Icon OpenAI Jupyter Notebook Icon Jupyter
Botify Project Image

Botify Grime builds on Botify by developing a playlist-based recommendation algorithm. Leveraging web-scraped genre spatial data from everynoiseatonce, the system visualizes genre data using 2D and 3D scatter plots, combines playlist and genre information, and computes weighted lists of parent and subgenres. It generates customized artist genres, fetches top tracks of related artists, and merges audio features with existing data. Through interactive user input, the system computes similarity scores for tracks based on desired feature values and importance levels, ultimately providing personalized music recommendations.

Python Icon Python Pandas Icon Pandas NumPy Icon NumPy Matplotlib Icon Matplotlib Seaborn Icon Seaborn Spotify API Icon Spotify Jupyter Notebook Icon Jupyter
Botify Grime Project Image

New Utility is a Flask-based web application that offers a comprehensive platform integrating OpenAI's functionalities. It enables text-to-speech, transcription and translation, image generation using DALL-E 2 & 3, a ChatGPT styled-interface with tuneable parameters and models, and a Q&A chat system that leverages user-uploaded documents, and images to enrich conversations.

Development paused as of March 30th, 2025. I simply don't have the time. :(

Python Icon Python Flask Icon Flask MySQL Icon MySQL HTML Icon HTML CSS Icon CSS JavaScript Icon JavaScript WebSockets Icon WebSockets Celery Icon Celery RabbitMQ Icon RabbitMQ NGINX Icon NGINX
New Utility Project Image

Hinterview is an application designed to aid job interviews by delivering real-time, personalized responses. The application boasts a user-friendly CLI equipped with customizable parameters and hotkey controls for maximum user control. During a Zoom interview, a hotkey is pressed to live record and transcribe audio. The transcription is then analyzed, and relevant document sections are retrieved based. The GPT model generates a response by considering these relevant sections, ensuring that the answers are contextually accurate and personalized.

Python Icon Python OpenAI API Icon OpenAI Blackhole Virtual Microphone Icon Blackhole VM
Hinterview Project Image

This ArcGIS StoryMap takes readers on a journey through the transformation of Paris over the centuries. It juxtaposes the city's portrayal in modern works like the 1995 film "La Haine" and Salim Bachi's 2008 story "Le Grand Frère" with historical texts such as Guillaume Apollinaire's 1913 poem "Zone" and François Rabelais's 1534 novel "Gargantua." By merging insightful spatial analysis with rich narratives spanning from the 16th to 21st centuries, the project uses an interactive narrative to illuminate how Paris's evolving urban landscape has profoundly shaped the stories and characters that emerge from it.

ArcGIS Icon ArcGIS French Flag French
ArcGIS - Paris Project Image

GoGuard is a Go-based VPN client that interfaces with Mullvad's WireGuard servers to provide secure and efficient VPN connections. The project offers features such as automatic server selection based on latency or geographic preferences, custom DNS configuration, key management, pre/post connection commands, and continuous connection monitoring.

Go Icon Go Wireguard Wireguard Mullvad Icon Mullvad Linux Icon Linux
Go Guard Image

Socks Toggle

Socks Toggle is an Automator app that simplifies toggling the SOCKS proxy settings on macOS. It allows users to easily enable or disable the SOCKS proxy for a specified Wi-Fi service, providing a convenient way to configure and use SOCKS proxies on their Mac. The script supports customizable settings for the Wi-Fi service, SOCKS proxy IP, and port, and it saves user-specified options for subsequent toggles.

Shell Script Icon Shell Script macOS Icon macOS
Socks Toggle Project Image

NYT Crossword Reset automates the process of identifying and resetting incomplete New York Times Crossword puzzles in a user's history. It uses Selenium WebDriver to interact with the NYT Crossword website, scanning the user's puzzle archive to find incomplete puzzles and clears them to ensure the history only reflects completed puzzles. The script offers three modes of operation (scan, fix, or both) and allows configuration of NYT credentials, number of months to scan back, and whether to run in headless browser mode.

Python Icon Python Selenium Icon Selenium
NYT Crossword Reset