A quiet corner of the internet where machine learning learns to sit still and look pretty — real‑time vision systems, gentle interfaces, and work still in metamorphosis.
I'm an emerging machine learning practitioner who spends most of her time teaching machines to see — traffic, faces, signals, the small everyday things most people scroll past.
Butterflymindspace is my working sketchbook: part portfolio, part proof that technical work doesn't have to look cold. Every project here starts as a rough, awkward larva of an idea and gets refined until it can fly.
Two portfolio pieces currently out of the chrysalis, built to be looked at as closely as they look at the world.
A live vision pipeline that watches the road the way I wish more systems did: gently, but precisely. Detects and tracks vehicles, estimates speed, reads license plates, and logs violations to a running dashboard.
A real‑time seizure detection model fed by simulated live EEG signal, streaming predictions as they happen rather than waiting for a notebook to finish running. Built to feel closer to a bedside monitor than a chart.
New work lands here as it's ready — check back, or say hello below and I'll let you know directly.
Detection, tracking, and recognition models built for real‑time footage — traffic, motion, and the small signals in between.
Deep learning on time‑series and biomedical signal data, streamed live instead of buried in a static notebook.
Presenting technical work — demos, dashboards, portfolios — so the craft is as legible as the code behind it.
“Intelligence, like a butterfly, should move as if it costs nothing — even when everything underneath is working very hard.” — Butterflymindspace
Open to machine learning roles, collaborations, and anyone who wants to talk computer vision until it gets weird.