jxnl.resume.md

Want to speak with me? I'm jason at jxnl.co

# First things first, Fizzbuzz
for i in range(1, 101): print "Fizz" * (not i % 3) + "Buzz" * (not i % 5) or i

# Or if you want... Simpson's Rule
def simpson(a, b, f, N):
    return (1.0 / 3.0) * (2 * (((b - a) / N) * sum(f(v) for v in [i * ((b - a) \
    / (2. * N)) for i in range(2 * N + 1)][1:2 * N + 1:2])) + (((b - a) / N) * \
    ((f(a) + f(b)) / 2.0 + sum (f(v * ((b - a) / N) + a) for v in xrange(1, N)))
    ))
    
I swear I write good code.

Work Experience

Hack the North -- Data Scientist

Sole data scientist working on the Hack the North team.

Exploratory data analysis and visulation to study the interests of hackathon participants to improve upcoming event.
Designed experiments on various biases that might exist in the application/judging process.
Created reports and summary statistics of the event and it's participants and gave suggestions to the directors.

NYU Global Institute of Public Health -- Research Intern (Current Position)

Supervisor: Dr. Rumi Chunara

Currently working on a paper discussing the use of machine learning models to find patterns of alcohol abuse on social media.

Exploratory analysis using Python and Gensim for topic modeling.
Built complex preprocessing and ingestion pipeline for machine learning with scikit-learn and Gensim.
Developed informative ipython notebooks that outline and document the body of work produced during the research project.
Using Amazon Mechanical Turk and crowdsource techniques to develop training data from raw twitter firehose.

Sysomos -- Data Scientist

Prototyped a proof of concept advertisement recommendation platform for offline targeted audience generation.
Developed extensible interfaces to our community detection and k-armed bandit service layer.
Improved clustering performance for twitter community detection on Sysomos MAP and Heartbeat.
Created MapReduce and Spark applications for Audience generation and various ad-hoc ETL.

Education

University of Waterloo 3B Honors B.Math, transfered from Mathematical Physics Computational Mathematics, Statistics Minor CO-OP

Applied Probability&Statistics, Linear Modeling, Data Visualization, Data Structures, Linear Algebra, Real Analysis, Computational Physics.

MOOCS (Coursera)

Machine Learning, from Andrew Ng
Data Science, from Bill Howe
Natural Language Processing, from Dan Jurafsky
Probabistic Graphical Models, from Daphne Koller

Projects

[Mark Sweep]http://devpost.com/software/mark-sweep)

Mark Sweep is a collection of services for automating Facebook group moderation. Prototype is currently being rewritten into a service oriented design for reuseability in other applications.

Won "Best use of Machine Learning" and placed Top20 at PennAppsX.
Weighed reservoir sampling was used for group based topic classification as a novel way to capture recency.
Topic/Troll detection done using using sklearn's SVC bag of word features, and watchwords.
Spam classification using hand engineered features.

Freelance

Provided various data services including web crawling, data analysis, and consulting.
Maintains 70% paid clientwork and 30% pro bono non profit.

Experiments

Itpy -- Lazy evaluated list processing with chained transformations such as streaming variance, groupbys, map, filter and more.
Reservoir -- Python module for uniform, exponential, and weighted reservoir sampling.
Bandit -- Java Implementation of various bandit algorithms.
Data Mining Canadian Goverment press releases (2002-2015) to uncover potential data journalism stories.

jxnl/jxnl.resume.md