Portfolio
Curated a privacy-conforming AML data set to support multiple university and NGO research initiatives.
Had my pull request to Luxor.jl accepted in order to add simple rotations to Turtle.
Wrote a program in Julia that uses group theory to quickly generate Sudoku puzzles and graph theory to solve them.
Designed and deployed a dashboard for easy exploration of spelling variations and entry errors in record data.
Successfully pitched to executives a bolt-on machine learning enhancement to a legacy system.
Developed an approximation to allow scaling of in-memory text clustering to hundreds of thousands of records on a single desktop machine with limited resources.
Resolved 500k+ Brazilian government records with our data in order to append new PII.
Gave a talk at the annual tech strategy conference on the theory behind and progress of a deep learning NLP enhancement to a legacy system.
Gave a tech talk on string distance metrics for fuzzy matching
Deployed a deep BiLSTM with custom model serving code as a bolt-on enhancement to a legacy system.
Reverse engineered an undocumented legacy fulfillment file to provide field-level diffs for a premier client.
Published on github a proof of concept for discreetly watermarking a database extract in a way that scales to 2^128 file recipients.
Hosted my first (real) Haskell project on Github, a small framework for speeding up fuzzy natural language string matching between two large sets of names.