Hello!

I'm Vithursan Thangarasa, a design-minded engineer focused on embedding artificial intelligence into applications that can transform our world in ways many of us can barely imagine

Get in touch vithursan "DOT" thangarasa "AT" gmail "DOT" com

Background

I'm currently a Machine Learning Engineer at Tesla working with a super talented team of engineers on future products and computer vision systems for Tesla's vehicles. As a graduate student advised by Dr. Graham W. Taylor, I am working towards an MASc at University of Guelph's Machine Learning Research Group (MLRG). I completed my BEng. in Engineering Systems and Computing, along with four awesome internships at Scotiabank, ON Semiconductor, Evertz Microsystems, and Jamdeo.

As a Machine Learning Researcher, I am interested in developing Continual/Lifelong Learning algorithms that allow Deep Neural Networks to continually evolve over time and dynamically adapt to unseen real-world environments at test time, and improve as they operate. I enjoy bridging the gap between machine learning research and engineering — combining my technical knowledge with my keen eye for design to create a beautiful product. My goal is to always build applications that are scalable and efficient under the hood while providing engaging, streamlined user experiences.

When I'm not in front of a computer screen, I'm probably playing basketball, working out at the gym, or drawing beautiful landscapes, architectures or cars with pencil.

Software Skills
Languages
  • Python
  • C
  • C++
  • MATLAB
  • HTML
ML Frameworks
  • PyTorch
  • TensorFlow
  • Scientific Python Stack
Models
  • CNN
  • RNN
  • LSTM
  • GAN
  • VAE
  • Seq2Seq
DevOps Tools
  • AWS
  • Kubernetes
  • Terraform
  • CloudFormation
  • Docker
  • Jira
  • Git
Hardware Skills
Languages
  • Verilog
  • VHDL
Design Tools
  • Xilinx ISE
  • Vivado HLS
  • GNU ARM Eclipse
  • Simulink
Hardware
  • Embedded Systems
  • Xilinx Zynq-7000
  • ARM Cortex-M
Research
Vithursan Thangarasa and Graham W. Taylor
British Machine Vision Conference (BMVC) 2018
We introduce Self-Paced Learning with Adaptive Deep Visual Embeddings (SPL-ADVisE), a novel end-to-end training protocol that unites self-paced learning (SPL) and deep metric learning (DML). Our architecture combines the strength of adaptive sampling, the efficiency of mini-batch online learning, and the flexibility of representation learning to form an effective self-paced strategy in an end-to-end DNN training protocol.
Curriculum Learning Diversity Joint Optimization CNN Embeddings PyTorch
Experience
May 2018 - Present
Machine Learning Engineer
Graduate Teaching Assistant
Deep Learning Data Scientist
Hardware and Systems Developer
Software Engineer, Video Compression
Mobile Application Developer, Android
View My Resume