Dylan Jayabahu

Resume

Download Resume

Get the full details of my experience, education, research publications, and professional certifications in PDF format.

Software Engineering @ UWaterloo AI/ML Research Full-Stack Development Publications in IEEE
View Resume

Research Publications

Conference Paper
2025

Deep Learning Model for Invasive Ductal Carcinoma Detection in Histopathology Images

Dylan Jayabahu (sole author)

IEEE CCECE 2025 IEEE Xplore

Introduces a novel computer-aided diagnosis (CAD) system for detecting Invasive Ductal Carcinoma (IDC) in histopathology images using advanced deep learning techniques. The system incorporates innovative sliding window-based heatmaps and a unique oversampling technique that extracts patches from homogeneous regions in whole-slide images, addressing class imbalance while maintaining biological integrity. Achieved state-of-the-art performance with 89.06% balanced accuracy and 86.68% F1-score, outperforming existing models in the literature.

Innovation: Novel sliding window heatmaps and homogeneous region oversampling
Performance: 89.06% balanced accuracy, 86.68% F1-score (SOTA)
Venue: IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)
Location: Vancouver, Canada (May 2025)
Impact: Clinical tool for pathologists with interpretable heatmap visualization
IEEE Xplore TechRxiv CCECE 2025
Journal Article
2024

Dataset for Real-World Human Action Detection Using FMCW mmWave Radar

Dylan Jayabahu, Parthipan Siva

CVIS 2024 JCVIS

Presents a real-world mmWave radar dataset for human action detection with applications in healthcare and home automation. Unlike existing research limited to controlled environments, this work captures natural activities across 28 residences, focusing on privacy-preserving sensing technology for aging-in-place monitoring. Developed and evaluated CNN-based models for action detection, demonstrating the challenges and potential of real-world radar-based human activity recognition systems.

Dataset: 28 homes, 458 sit-down and 454 stand-up actions
Technology: TI IWR6843AOP FMCW mmWave sensor with 3D point clouds
Model: 3-layer CNN with sliding window detection approach
Presentation: Presented at CVIS 2024, University of Waterloo
Innovation: First real-world dataset vs. controlled simulations
JCVIS arXiv CVIS 2024

Featured Projects

Professional Certificates

Showing 6 certificates
IBM AI Engineering with Python, PyTorch & TensorFlow Specialization
IBM

AI Engineering with Python, PyTorch & TensorFlow

Achieved: October 2025
Machine Learning Deep Learning PyTorch TensorFlow Keras LLMs
View Certificate Github
IBM Generative AI Engineering with LLMs Specialization Certificate
IBM

Generative AI Engineering with LLMs Specialization

Achieved: October 2025
LLMs NLP Transformers Fine-Tuning PyTorch RAG LangChain
View Certificate Github
IBM deep learning with pytorch keras and tensorflow professional certificate
IBM

Deep Learning with Pytorch, Keras and Tensorflow Specialization

Achieved: August 2025
Python Tensorflow Pytorch Keras DL
View Certificate Github
Data structures and algorithms Professional Certificate
GeorgiaTechX

Data Structures & Algorithms

Achieved: August 2025
Java OOP ADTs
View Certificate Github
Data science machine learning certificate
HarvardX

Data Science: Machine Learning

Course Completed: February 2025
ML Algos R Caret (R)
View Certificate Github
computer science for ai profesionla certificate
HarvardX

Computer Science for Artificial Intelligence

Achieved: January 2025
Python NLP ML SQL C
View Certificate