Working on risk analysis tools that make order fulfillment safer for Shopify merchants.

Researched adaptive user interfaces and applying machine learning to build user models. TA for Advanced Operating Systems, Artificial Intelligence, and Data Structures for Electrical Engineers.

I tested the scalability of schemes for encoding structured representations with populations of neurons. What this means, in practice, is that I mapped vocabulary items from the WordNet lexical database of English into neural firing rates, which could then be queried with similarly encoded concepts representing lexical relationships to produce an output pattern representing an associated vocabulary item. To use a concrete example, the word 'computer'​ in WordNet has the hypernym 'machine'​… Show more I tested the scalability of schemes for encoding structured representations with populations of neurons. What this means, in practice, is that I mapped vocabulary items from the WordNet lexical database of English into neural firing rates, which could then be queried with similarly encoded concepts representing lexical relationships to produce an output pattern representing an associated vocabulary item. To use a concrete example, the word 'computer'​ in WordNet has the hypernym 'machine'​ (because a computer is a machine), so the neural network would be given a representation of the word 'computer'​ and of the relationship 'hypernym'​ and would produce a representation of the word 'machine'​. It should be noted that the representational schemes and encodings used in this work were based on existing work by Tony Plate on holographic reduced representations (HRR) as well as previous work by Prof. Eliasmith's research group. My own contribution was to explore the effects that recursively binding representations together into a hierarchical structure (such that the 'machine'​ representation in the previous example could be queried with the hypernymy relationship to produce the representation for 'device'​ and so on) would have on representational accuracy.E. Crawford, M. Gingerich, and C. Eliasmith (2013). “Biologically Plausible, Human-Scale Knowledge Representation” in 35th Annual Conference of the Cognitive Science Society (pp. 412-417). Cognitive Science Society.I also worked on some other stuff, most of which is now part of the book "How to Build a Brain" by Chris Eliasmith (Oxford Press, 2013). Show less

Implementation of remote sensing algorithms.

CREZ creates software to collect and manage sports video and statistical data.I managed the stabilization of the client application for its version 3.1 release.

I worked with Autodesk's User Experience group to create user interfaces that were easy to learn and to work with. My main role on the team was to develop software prototypes of proposed or experimental interface designs.