About

I started in neuroscience as an undergraduate at York University in Toronto and have stayed in some form of brain or language research ever since. After the BSc I went to the University of Toronto for an MSc and then a PhD in Speech-Language Pathology, both supervised by Yana Yunusova at the Sunnybrook Research Institute. My dissertation was on bulbar amyotrophic lateral sclerosis, the variant of ALS that hits the motor systems used in speech first. I tried to characterize the breakdown across modalities, combining detailed acoustic and kinematic measurements of speech with structural MRI and, ultimately, the neuropathology of patients who had passed away. The work was slow, the datasets were small, and the question of what the brain of someone whose speech is being progressively dismantled actually looks like, in life and after, is one I still find genuinely compelling.

After Toronto I moved to Philadelphia for a postdoctoral fellowship with Murray Grossman at the University of Pennsylvania's Frontotemporal Degeneration Center. The Penn FTDC is one of the larger groups in the world studying frontotemporal degeneration, and the lab's particular strength is treating language as a window onto neurodegenerative pathology. My job was to apply the digital speech analysis methods I had developed for ALS to a wider set of conditions. The patient populations I worked on there included primary progressive aphasia in its several variants, behavioural-variant frontotemporal dementia, ALS-FTD spectrum disorders, Alzheimer's, Lewy body dementias, and asymptomatic carriers of pathogenic genetic variants.

My first-author work from the postdoc sits mostly in the FTD-Alzheimer's and PPA literatures. The 2024 paper in Neurology on asymptomatic carriers of pathogenic FTD variants got the most attention. We found measurable speech changes years before clinical onset in those carriers, and the National Institute on Aging covered the result. Other first-author papers from the postdoc are on lexical and acoustic features that relate to Alzheimer's pathology, on apraxia of speech in nonfluent/agrammatic PPA, and on using machine learning over natural speech to detect Alzheimer's pathology in FTD patients. The full list is on the research page.

I left academia in the summer of 2024 and joined BLDS LLC, a statistics and economics consulting firm that has been doing quantitative analysis of discrimination since the 1980s. BLDS is also the home of SolasAI, the firm's platform for auditing machine learning models for bias and disparate impact. My current work is a mix of conventional fair-lending statistical analyses and applied research on systems that involve language. Recent projects include a retrieval-augmented generation system built for a client, an embedding-based analysis of how an employment-screening model matches candidate resumes to job descriptions, and a small language model I'm training to detect discriminatory speech turn by turn, so that each phrase can be flagged or denied at the moment of generation.