About
Thejaswini co-founded Quantis Sphere and leads its engineering and data work — the technical architecture behind the firm's AI and data systems, and the research discipline it brings to client engagements: reproducible method, stated limitations, and evidence for every claim.
She is a Project Engineer at IHFC, the I-Hub Foundation for Cobotics, on a ₹3 crore funded programme developing an affordable AI-driven bionic hand, and a direct PhD candidate in Computer Science and Engineering. Her research in biomedical signal processing and deep learning has produced three published patents and Scopus-indexed publications.
Experience
Project EngineerCurrent
IHFC — I-Hub Foundation for Cobotics
Works on a ₹3 crore funded research programme developing an affordable, AI-driven bionic hand — applying deep learning and biomedical signal processing to gesture recognition and prosthetic control.
Co-Founder & Lead Technical EngineerCurrent
Quantis Sphere LLP
Leads engineering and data architecture across the firm's client systems, and the applied AI work behind them.
Education
Direct Ph.D. (pursuing) in Computer Science and Engineering
Karunya Institute of Technology (Deemed University) · 2026 – Present
B.Tech in Computer Science and Engineering
Karunya Institute of Technology (Deemed University) · 2025
Skills
Data Science & Quantitative Analysis
Machine Learning & Deep Learning
Signal Processing & Research
Engineering
Research interests
Patents(3)
System and Method for the Classification of Prosthetic Hand Movements Using Deep Learning Techniques
Indian Patent Office · 2026
An automated framework that converts dual-modality biosensor data (22-channel glove-sensor recordings and angle measurements) into time-frequency spectrograms to classify prosthetic hand movements using deep learning.
Method and System for Multiple Hand Movement Using Deep Learning to Aid Rehabilitation in Prosthesis
Indian Patent Office · 2025
A system that classifies hand movements from EMG signals using Short-Time Fourier Transform (STFT) spectrograms and a DenseNet-101 deep learning model to aid prosthesis rehabilitation.
BCI Based EEG Motor Imagery Classification Using Spectrogram Images and EfficientNet-B0 Network
Indian Patent Office
A brain-computer-interface system that classifies EEG motor-imagery signals from spectrogram images using an EfficientNet-B0 transfer-learning model for neuro-rehabilitation applications.
Journal Publications(2)
Hand Gesture Recognition Using Quantum Inspired Neural Networks with sEMG
SN Computer Science · 2026
Enhanced EMG-based Gesture Recognition using Hybrid CNN-BiLSTM Architecture with Channel Attention
Biomedical and Pharmacology Journal · 2025
Conference Papers(3)
Neural-Muscular Coupling Analysis for Upper Limb Gesture Classification: A Multimodal EEG-EMG Fusion Framework
International Conference on Emerging Systems and Intelligent Computing · 2026
Enhanced Bilateral EMG Classification Through Integration of Graph Convolution, Transformer, and BiGRU Networks
International Conference on Emerging Trends in Industry 4.0 · 2025
An Effective Stock Market Prediction using an Advanced Machine Learning Algorithm and Emotional Analysis
3rd International Conference on Applied Artificial Intelligence · 2024
Projects(4)
Quantum-Inspired Neural Networks for Hand Gesture Recognition
A quantum-inspired neural framework for multimodal gesture recognition on sEMG and CyberGlove signals (NINAPRO DB1: 52 gestures, 10 subjects). Q-RNN, Q-CNN and a Quantum Ensemble Design reaching 77.5% accuracy, with a full preprocessing pipeline — segmentation, statistical features, PCA and quantum-inspired mapping.
Multimodal EEG–EMG Fusion for Upper-Limb Gesture Classification
A fusion framework integrating EEG and EMG signals for neural-muscular coupling analysis; a stacking-classifier ensemble on the Mendeley dataset reached 94% accuracy with statistical significance (p < 0.01).
EMG-Based Gesture Recognition with Channel Attention
A hybrid CNN–BiLSTM architecture with channel attention for EMG classification, evaluated on accuracy, precision, recall and F1 against baseline models.
Motor-Imagery EEG Classification
Spectrogram analysis with EfficientNetB0 transfer learning for motor-imagery classification, with feature extraction, cross-validation and statistical analysis.
Certifications(4)
Work directly with the people who do the work.
A 20-minute call. You get a written scope and a fixed price — and an honest answer if we're not the right people for it.