EEG deep learning
Learning discriminative representations from noisy, non-stationary EEG, using architectures and training regimes suited to small, subject-variable biomedical datasets.
Decoding intention from EEG.
PhD research in biomedical signal processing and brain-computer interfaces at NIT Calicut: deep-learning models for motor imagery classification, built toward clinically viable neurotech for rehabilitation.
Focus areas.
Motor imagery classification
Classifying imagined movement from brain activity: the core decoding problem behind BCIs for rehabilitation and assistive control.
Biomedical signal pipelines
End-to-end pipelines: acquisition, artifact handling, feature learning and evaluation protocols designed for reproducibility and clinical relevance.
Publications.
My publication record in BCI is focused and emerging: one peer-reviewed paper, active PhD work, and a decade of adjacent engineering practice.
TriMiX: An Interpretable Machine Learning Pipeline for Tri-Class Motor Imagery EEG Classification for Clinical Brain-Computer Interfaces
Most motor-imagery BCIs recognise only two mental commands and never model a resting state, so a prosthetic can activate during idle moments, undermining safety and trust. TriMiX adds that third "rest" class and pairs interpretable neurophysiological EEG features (Filter Bank Common Spatial Patterns and bandpower) with statistical feature selection and an XGBoost classifier. On the 109-subject PhysioNet EEG dataset it reached 96.26% subject-wise accuracy (77.9% leave-one-subject-out cross-validation), a step toward safer asynchronous control for clinical brain-computer interfaces.
Read on IEEE Xplore →Combined Utility and Adaptive Residence Time Based Network Selection Algorithm for 4G Wireless Networks
Received the Best Paper Award at the 3rd International Conference on Multimedia Technology and published as a Springer book chapter in Lecture Notes in Electrical Engineering, Springer.
For research teams.
I bring doctoral-level BCI work, ten years of teaching that sharpened how I communicate complex systems, and hands-on telecom engineering. Open to Research Scientist and Research Associate roles in biomedical AI: industry or academia.
Remote-first. For on-site roles: Thiruvananthapuram, Kerala, India.
Start a conversation
Snapshot for hiring teams
- PhD Scholar: ECE, NIT Calicut · biomedical signal processing & BCI
- M.Tech: Communication Systems, SRM University, Chennai
- B.Tech: ECE, UCE Kariavattom, University of Kerala
- Former Assistant Professor: APJ Abdul Kalam Technological University, 10+ years
- Industry: telecom engineering; award-winning 4G networks research
- Recognition: 4× IEEE awards · Best Paper · quoted in Forbes · featured in The Times of India
- Communication: 481 expert sessions; can explain the work to any room
Open work.
GitHub
Theory Meets Practice
Open, hands-on notebooks bridging theory and implementation, e.g., eigen-decomposition and SVD applied to image compression.
View repository →
Curated
Research tools bundle
The AI-tools stack I use and teach: discovery, writing, visualisation, reference management and analysis.
Open the bundle →Hiring for biomedical AI, EEG or BCI?
I'm actively exploring research roles where clinically meaningful neurotech is the mission. Tell me what your team is building.
Contact Shankar