Home

I’m a Research Engineer at Borealis AI Montreal working on foundation models and LLMs for finance. I have 6+ years of full-time work experience in applied AI.

Previously, I was a Research Engineer at Samsung AI Center Toronto working with Dr. Alex Levinshtein and Prof. Allan Jepson on computer vision, specifically, burst photography, neural implicit models, image enhancement and synthesis.
Earlier, I was a Software Engineer developing behavior-based malware classifiers using machine learning for Norton AntiVirus at Symantec Corporation in Pune, India.

I’m currently interested in fundamental research on the generalization, calibration and robustness of deep neural networks, and applied AI for healthcare and medicine.

To discuss collaborations and opportunities please reach out to me by email -
vin [dot] bhaskara [at] gmail [dot] com

Education

M.Sc. Applied Computing (Computer Vision) 
Department of Computer Science, University of Toronto 2018 - 2020
Grade: A+ [4.0/4.0], Vector Scholar in AI  
Academic Supervisor: Prof. Roger Grosse  
   
B.Tech. Engineering Physics 
Indian Institute of Technology Guwahati (IIT Guwahati) 2012 - 2016
Institute Silver Medalist, IQC Research Visitor  

Patents

May
2022
   Unsupervised Super-Resolution Training Data Construction
Haicheng Wang, Xinyu Sun, Vin Bhaskara, Stavros Tsogkas, Allan Jepson, Alex Levinshtein
US Patent Application 17/512,312

Patent Cite

Publications

Machine Learning

Jan
2022
   GraN-GAN: Piecewise Gradient Normalization for Generative Adversarial Networks
Vin Bhaskara*, Tristan Aumentado-Armstrong*, Allan Jepson, Alex Levinshtein
Winter Conference on Applications of Computer Vision (WACV 2022)

Paper Arxiv Poster Slides Video Cite
     
Jan
2021
   Efficient Super-Resolution Using MobileNetV3
Haicheng Wang*, Vin Bhaskara*, Alex Levinshtein*, Stavros Tsogkas, Allan Jepson
European Conference on Computer Vision (ECCV 2020) Workshops

Paper PDF Cite

Quantum Information

May
2022
   Generalized Entanglement Measure for Continuous-Variable Systems
Nibedita Swain*, Vin Bhaskara*, Prasanta K. Panigrahi
Physical Review A 105, 052441

Paper PDF Arxiv Cite
     
Mar
2017
   Generalized Entanglement Measure for Multiparticle Pure States in Arbitrary Dimensions
Vin Bhaskara, Prasanta K. Panigrahi
Quantum Information Processing, Volume 16, Article number: 118

Paper PDF Arxiv Cite
     
Mar
2017
   Implementing Bragg Mirrors in a Hollow-Core Photonic-Crystal Fiber
Jeremy Flannery, Golam Bappi, Vin Bhaskara, Omar Alshehri, Michal Bajcsy
Optical Materials Express, Volume 7, Issue 4, pp. 1198-1210

Paper Cite
     
Sep
2016
   Mesoscale Cavities in Hollow-Core Waveguides for Quantum Optics with Atomic Ensembles
C.M. Haapamaki, J. Flannery, G. Bappi, R. Al Maruf, Vin Bhaskara, O. Alshehri, T. Yoon, M. Bajcsy
Nanophotonics, Volume 5, No. 3, pp. 392-408

Paper Cite
     

(* denotes equal contribution)

Preprints

Apr
2020
   Part-based Auxiliary Objectives with No Extra Labels for Robust Single-Shot Object Detection
Vin Bhaskara, Stavros Tsogkas, Kosta Derpanis, Alex Levinshtein
DOI: 10.13140/RG.2.2.10079.47521

Link PDF Cite
     
May
2019
   Exploiting Uncertainty of Loss Landscape for Stochastic Optimization
Vin Bhaskara, Sneha Desai
arXiv:1905.13200 [cs.LG]

Arxiv Cite
     
Apr
2019
   Risk Prediction in the General Internal Medicine Ward at St. Michael's Hospital
Vin Bhaskara, Yingying Fu, Sindhu Gowda
DOI: 10.13140/RG.2.2.27695.55205

Link PDF Cite
     
Jul
2018
   Emulating Malware Authors for Proactive Protection using GANs over File Behaviors
Vin Bhaskara, Debanjan Bhattacharyya
arXiv:1807.07525 [stat.ML]

Arxiv Cite