Michelle Fullwood

About

Summary

Natural Language Processing Scientist with strong background in linguistics and math
10+ years of experience in Python and web development

Education

Ph.D. in Linguistics, Massachusetts Institute of Technology (2010-2018)

Computational & experimental research on acquisition of Arabic morphology
Dissertation: Biases in Segmenting Non-concatenative Morphology
Coursework in linguistics, machine learning and Bayesian techniques
Teaching assistant in introductory linguistics and phonology

B.A. in Linguistics and Mathematics, Cornell University (2000-2004)

Graduated magna cum laude and with distinction in all subjects, 4.0 GPA
Study abroad, Fall 2003: Budapest Semesters in Mathematics

MOOCs from edX, Coursera, FutureLearn, etc. (ongoing)

Constantly improving my data science, deep learning, and programming skills

Experience

Natural Language Processing Scientist, Spirit AI (2017–present, remote)

Build pipeline of preprocessors and classifiers for detecting toxicity in text
Research and implement NLU algorithms for chatbot–like product
Build in-house tools for annotating and managing data

Consultant, Vista Higher Learning (2015–2017, Boston MA)

Advised client on architectural design and data collection for Spanish and French speech recognizers for second language learners
Analyzed data to better understand corpus characteristics and recognizer performance

Web Developer, Imperial Consulting (2008-2010, Boston MA)

Developed custom web applications, both front and back end, in Python, Django and jQuery, for clients in education, biomedical and finance industries

External Consultant, Centre for Strategic Infocomm Technologies (2008–2009, Singapore)

Advised client on evaluation procedures in speech and natural language processing

R&D Engineer, later Consultant, Centre for Strategic Infocomm Technologies (2004–2008, Singapore)

Researched techniques and built engines for speech recognition, language and speaker identification, and cross-language information retrieval
Managed projects to evaluate and acquire systems

Summer Intern, PARC (2004, Palo Alto CA)

Engineered a computational Lexical Functional Grammar to cover the basic sentence constructions of Modern Standard Arabic
Built a root-and-pattern-based finite state Arabic morphological analyzer

Skills

Programming:

General: Python, JavaScript, git
NLP/DS: SpaCy, PyTorch, scikit-learn, gensim, Pandas, Spark, R, SQL
Web: Django, Flask, VueJS, jQuery

Languages: English (native), French (advanced), Mandarin (advanced), Arabic (intermediate), Japanese (intermediate), Hungarian (elementary), Hieroglyphic Egyptian (elementary)

Awards

National Science Foundation Graduate Research Fellowship (2011)
Merrill Presidential Scholar, Cornell University (2004)
Achievement Award in Intermediate Arabic, Cornell University (2004)
Achievement Award in Elementary Arabic, Cornell University (2002)
College Scholar, Cornell University (2001)
Dean's Scholar, Cornell University (2000)
Dean's List (2000-2004)
Public Service Commission (Singapore) Overseas Merit Scholarship (2000)

Other

Python community

Boston Python co-organizer
Volunteer tutor and event organizer for PyLadies Boston (2013-2016)
Speaker at PyCon 2015, 2017, PyData NYC 2014, numerous occasions at Boston Python and PyLadies Boston

Linguistic Streetmap of Singapore (2015)

Used machine learning (scikit-learn) and mapping technologies to create a visualization of linguistic origins of Singapore streetnames
Link

Fuzzy Arabic Dictionary (2014, personal)

Developed a novel Arabic dictionary that can be queried via transliteration, rather than needing to know the exact orthography,
Mash-up of the Yamli transliteration service and Buckwalter Arabic Morphological Analyzer
Link