Matt Dennewitz
10
Recent Work
-
Wiggle Hunt,
where users can search new and used modular synth inventories
from stores around the world
-
Baseball Prospectus,
for Statcast data architecture
Job Work
By day, I work at Conde Nast. Recently I was the VP of Product for Pitchfork,
Engineering Director for both Pitchfork and WIRED, and I built the first version
of them.us. Now, I am the Engineering Directory for CN's new User Utility group.
Selections from GitHub
Here's my GitHub profile,
which has a lot of baseball data code, api wrappers for a couple of music sites,
and some other odds & ends.
-
playlist-to-vec,
a word2vec-based music recommendation engine powered by analyzing Spotify playlists.
-
hash-ring-ctypes,
a fast ctypes-based wrapper for libhashring
-
django-active-login-required,
a view decorator that ensures authenticated users are also still marked as active
-
python-acoustid-api,
a light wrapper around AcoustID's lookup APIs
-
baseball-pagerank,
an experiment in surfacing research through graph centrality
-
rsd,
which builds a database of record stores from Record Store Day participant data
-
mlbam-utils,
which facilitates customizable MLBAM event log ingestion
-
homers,
an app that ingests live MLBAM play-by-play logs and emits home runs
through a public-facing JSON API
-
retrosheet-pitch-sequences,
for extracting ball-strike counts seen in a pitch sequence (i.e., pass-through counts)
from Retrosheet pitch sequences
-
retrosheet-queries,
a cookbook of retrosheet recipes for those interested in exploring sabermetrics
-
baseball-projection-schematics, a toolkit for mapping baseball player projection CSV sources, which come in all shapes and sizes, to a unified format
-
mlb-normalize-player-ids, for combining and normalizing player identification tags from disparate sources
Recent Talks
-
MIT Sloan Sports Analytics Conference, 2017, where I gave a lecture on
the ins and outs of web scraping. In this session, we built a composite ranking
of MLB prospects from different sites. In doing so, we looked at scraping dirty HTML,
JSON APIs, looking for structured metadata, XPATH, and more.