Hey, I’m Visare.
I’m an ML engineer working with signals from muscles, brains, and sensors.
I build end-to-end systems—from raw neural and biosignal streams to training, evaluation, and model serving.
I care about work that’s measurable, reproducible, and shipped.
USC ISI / Alexandria
USC COOR Lab
BrainSightAI
HMA Lab, IIIT Sri CityA few things I’m proud of:
- Published three IEEE papers on classifying daily activities from muscle signals — two in Rome, one in the US — carrying each one from problem statement to publication.
- Ran the two-year collection behind EMAHA-DB4–DB7: recruited and recorded 70 subjects, then released all four datasets publicly on Kaggle and Harvard Dataverse for other labs to build on.
- Found and fixed a signal-processing bug in CuPy —
bilinearwas quietly sensitive to leading zeros — plus a missing API doc; both merged upstream. - Caught a denoising failure at BrainSightAI that SNR and RMSE both passed — visible only on rendered FA maps — and changed the production default on that evidence.
A collection of experiments
Selected work
Things I’ve built, measured, and learned from.
2026
Paper Copilot
A literature-review companion. Entirely offline.
2025
Neural Decoding
From neural activity to movement.
2025
Reading Muscle Signals
Finding the moment a muscle switches on.
2025
Predictive Maintenance
Listening to machines before they fail.
2024
The Shape of a Voice
Learning to hear how we sing.
2025
Walking on Shifting Ground
A quadruped, loose terrain, and timed vibration.
Notes from the work
Writing
Research notes, practical guides, and things worth sharing.
Get in touch
Let’s make something useful.
Open to full-time ML / signal-processing roles for 2026, especially sensor ML, neurotechnology, and ML infrastructure.
vvenna@usc.edu