I am a Freelance Research Programmer with seven years of lab experience in the fields of Linguistics, Computer Science, and Psychology. I possess the knowledge and skills to significantly increase the research output of your lab by resolving technical bottlenecks. I specialize in the design, implementation, and distribution of web experiments, as well as other key services such as python scripting, data analysis, stimuli generation, and training neural networks.
Representative Projects
Open-source web app for distributing web experiments
Details / Github
Designed and built a full-stack web application that interfaces with the Mechanical Turk API to automate the process of releasing web experiments with multiple lists/conditions. By distributing the tasks in "micro-batches" of nine subjects each, it takes advantage of mTurk's 20% small batch discount while also insuring that the exact number of responses for each condition are collected. Designed for use with IbexFarm platform, but easily adjustable to any web experiment.
Check out the code for the project here
Custom experiment studying the ways people use predictive text systems
Details / Demo
Implemented a novel experimental paradigm in which subjects type out sequences of phrases from audio prompts using a custom word-prediction/auto-completion interface. On the back-end, I leveraged rapid part of speech tagging and output from a language model to create conditions in which the system provided either exclusively grammatical, exclusively ungrammatical, or purely nonsensical predictions and completions, to test the salience of grammaticality for users of such systems.
Try the experiment out here. (Desktop only. May take a moment to load.)
Mechanical Turk task designed to farm user intuitions about adjectives
Details / Demo
Built a custom Mechanical Turk task to collect human judgments on the intuitive ordering of adjectives along a given scale (for instance, on the scale of "quality": "bad" < "okay" < "good"). Responses were then used as training data for a neural network learning a similar task.
Try the experiment out here. (May take a moment to load.)
