I build machine-learning and computer-vision pipelines that turn ordinary video into quantitative motor biomarkers — detecting and classifying gait abnormality in cerebral palsy, grounded in clinical biomechanics.
I work at the seam between clinical biomechanics and brain-inspired machine learning.
An MS researcher in Medical Science and Technology at IIT Madras, in the Computational Neuroscience Lab under Prof. V. Srinivas Chakravarthy. My research extracts biomechanical motor biomarkers from video — knee asymmetry, trunk lean, pelvis sway — and uses them to detect and classify gait abnormality in children with cerebral palsy.
The throughline across everything I build is one clean pipeline: raw movement → pose → signal → decision, with oscillator-based neural networks (DONN / ResHopf) as a recurring tool.
How I got here.
Things I’ve built.
A full clinical web application for cerebral-palsy gait assessment: patient management across GMFCS / MACS / CFCS / EDACS, session-based video upload, frontal & sagittal analysis with threshold-based metric cards, and longitudinal tracking across visits.
An ST-GCN pipeline for CP-vs-typical classification from BlazePose keypoints, modelling the body as a spatio-temporal graph of joints over the gait cycle.
An oscillatory deep network built on Hopf-oscillator dynamics — reused across gait, surface EMG and upper-limb kinematics.
Tri-branch ST-GCN / CNN / Hopf classifying dyskinetic CP vs typically-developing reaching on the KU Leuven DCP dataset (Vanmechelen et al. 2024).
Musculoskeletal locomotion in MyoSuite / MuJoCo with PPO and SAC, a curriculum scheduler, a CP-impairment injection model, and a stability gate.
A DONN-based GMFCS severity classifier on the Stanford mobile-gaitlab dataset (Kidziński et al. 2020), with automatic view correction.
DONN-ResHopf classification of elbow isometric contractions on a Mendeley HD-EMG dataset.
The same system-design instinct, applied for fun.
Architecture for a full chess platform.
A hands-on music-production tool.
KP / Vedic charts with a real ephemeris.
Batting-biomechanics analysis; writing poetry.
Research collaboration, clinical-movement data, a project, or just to talk gait and ML — my inbox is open.