Rohith Shinoj Kumar

Hi! I'm a Computer Science and Engineering graduate from NITK, and I'm passionate about building AI systems that work reliably in clinical settings. I integrate ideas from control theory and deep learning to design models that stay stable and reliable, particularly in noisy sensor signals.

This interest comes from my work in medical imaging research and my current role as a Scientist at the Centre for Development of Telematics. I enjoy tackling problems that require both engineering research and mathematical foundations with a mission to create AI tools that are trustworthy, efficient, and ready for real-world deployment, especially on devices with limited resources.

Rohith Photo

Research

Paper thumbnail

H-Infinity Filter Enhanced CNN-LSTM for Arrhythmia Detection from Heart Sound Recordings

Rohith Shinoj Kumar, Rushdeep Dinda, Aditya Tyagi, Annappa B, Naveen Kumar M R

In Press at 2025 13th IEEE International Conference on Systems Engineering and Technology (ICSET)

Preprint Code

Paper thumbnail

A Deep Learning Framework for Automated and Consistent Ejection Fraction Quantification in Echocardiography

Chaitanya M, Rohith Shinoj Kumar, Jeny Rajan

Under review at International Journal of Biomedical Imaging

Author-submitted manucript available upon request

Paper thumbnail

Automated Rib Segmentation in Chest X-rays Using ThoraxSegNet: Enhancing Pulmonary Disease Detection and Analysis

Poornanand Naik, Rohith Shinoj Kumar, M P Singh

Under review at Engineering Research Express (ERX)

Author-submitted manucript available upon request

Paper thumbnail

DL-ARP: A Deep Learning based framework for dynamic Detection and Mitigation of ARP Spoofing attacks

Harshith Puram*, Rohith Shinoj Kumar*, BR Chandavarkar

2023 14th IEEE International Conference on Computing Communication and Networking Technologies (ICCCNT)

Paper Code

Industrial Experience

Centre for Development of Telematics (2025) — Designed and implemented performance-critical modules and for data analysis and deep packet inspection (DPI) within a national-scale firewall system. Implemented ONNX quantization and optimizations to work under throughput of over 700,000 packets/sec at sub-microsecond latency in live deployment

Indian Council of Agricultural Research (Under MoU with Vision & Image Processing Lab - NITK) Worked as a team of 4 to build an end-to-end system for detecting and classifying diseases from mobile-camera images of pomegranate leaves and fruit. My work focused on designing data-centric augmentation pipelines for generalization and ONNX optimizations for fast, device-agnostic inference.

Accenture Inc (2024) — Built time-series ensemble Prophet model for dynamic pricing, achieving over 92% accuracy in demand forecasting from large-scale customer analytics data.