CVPDF | JSON
Engineer with experience in machine learning and backend development.
Backend coding for a educational web app on Google App Engine (Python, JS).
Developed data model for a new progress mechanic that increased time spent on site by 3.5% ($p < 0.0005$).
Built a pipeline to compute and display practice task time estimates using BigQuery.
Built a new site for African-American Studies Dept. with designer Remeike Forbes (PHP/WP).
Enriched UX with a responsive WordPress theme.
Saved client labor by writing a plugin for importing academic publications.
Created Loom, a library/DSL for patterning audiovisual events (Java).
Improved Processing's frame-based timing (>30 ms) accuracy) to <1 ms by writing a soft real-time scheduler.
Added a track recommendation feature to DJ software based on audio similarity (C++).
Enabled efficient feature extraction for user libraries.
MetaLAB @ Harvard
Developed Paper Machines, Firefox extension for NLP/dataviz (Python, D3).
Offered insight into users' text corpora via topic modeling, NER, geoparsing.
Built a keyboard harmony trainer (JS).
App used by all intro music theory courses the next year.
Conducted experiments on genre classification and sample identification in audio data (Python).
Compared low-level audio features with CNN activations as topic model inputs.
Visualizing state transition graphs for servant RESTful APIs (Haskell).
New features, refactoring, and functional tests for 70k-user image-sharing app (Rails).
Visualization/sonification for topic models of audio (Python, Elm).
Presented at Global Digital Humanities 2015.
Analysis and visualization of Original Hip-Hop Lyrics Archive (Python, R, D3, Ember).
Shown at NIPS 2013 Topic Modeling Workshop.
Continued maintenance after GSoC.
Broadened compatibility by switching to Jython.
Wrote a new server for multithreaded NLP (Scala/Play/Akka).
Viz/mashup of Last.FM and Twitter API data (JRuby).
Tracked musicians' relationships across social platforms.