Welcome

I am a Speech ML Engineer and Applied Scientist specialising in speech recognition for clinical populations. With a PhD in Cognitive Neuroscience from University College London, I build AI systems that work for people whose speech doesn't fit the patterns mainstream technology was designed for — including speakers with dysarthria, aphasia, and other neurological speech disorders.

My recent work includes fine-tuning OpenAI's Whisper for dysarthric speech recognition, achieving a 47% reduction in word error rate using parameter-efficient adaptation (LoRA). The model and code are publicly available.

Previously, I built NUVA — a CE-marked Class II medical device for speech-driven therapeutic interventions in aphasia using on-device speech recognition — and published at INTERSPEECH on speech technology for clinical applications.

Explore my projects, read the blog, or learn more about me.