Skills

Programming Languages

Python and C++

Machine Learning Frameworks

PyTorch, Keras, Scikit-Learn, Tensorflow, Apache MXNet, ONNXRuntime

Tools

Docker, Numpy, Pandas, Scipy Stack, OpenCV, Pillow, Matplotlib, Seaborn, flask, nginx, uwsgi, boto3, git, Jira

Cloud and MLOps

Amazon AWS: EC2, S3, Batch, SQS, SNS, EFS, VPC, ECS, ECR, Lambda, CloudFormation, CloudWatch

Knowledge Areas

Machine Learning, Deep Learning, Computer Vision, Medical Imaging, Reinforcement Learning

Core-CS

Algorithms, Data Structures, Databases, Operating Systems, Linux, Probability, Statistics, Linear Algebra

Experience

 
 
 
 
 

Engineering Manager, Machine Learning (NLP)

BenchSci

Jan 2023 – Present Vancouver, BC Canada
 
 
 
 
 

Director of Engineering (Innovation Projects)

Xtract AI

Apr 2022 – Jan 2023 Vancouver, BC Canada
 
 
 
 
 

Engineering Manager (Innovation Projects)

Xtract AI

Jul 2021 – Apr 2022 Vancouver, BC Canada
 
 
 
 
 

Senior Machine Learning Engineer

Xtract AI

Sep 2020 – Jul 2021 Vancouver, BC Canada
 
 
 
 
 

Machine Learning Engineer

MetaOptima Technology Inc.

Sep 2019 – Sep 2020 Vancouver, BC Canada
 
 
 
 
 

Machine Learning Researcher

Division of Neurosurgery at Vancouver General Hospital

May 2019 – Aug 2019 Vancouver, BC Canada
 
 
 
 
 

Software Engineer (NLP) (Consulting)

Pluto Health Innovations (PlutoHI)

Apr 2018 – Aug 2018 Redwood City, CA USA (Remote)
 
 
 
 
 

Research Assistant in Machine Learning and Medical Imaging

Medical Image Analysis Laboratory at Simon Fraser University

Jan 2018 – May 2019 Burnaby, BC Canada
 
 
 
 
 

Machine Learning Engineer

RadSupport Inc.

Apr 2016 – Apr 2017 Mountain View, CA USA (Remote)
 
 
 
 
 

Other Experiences

Dec 2012 – Apr 2017
Complete list of previous experiences not listed here available on LinkedIn: https://www.linkedin.com/in/trane293/

Awards

NSERC-CREATE Bioinformatics Scholarship

Natural Sciences and Engineering Research Council of Canada Sep 2018 – Aug 2019
  • Scholarship award worth CA$22,000, over a period of one year for undertaking research and courses in bioinformatics. The program is run jointly by UBC and SFU.
  • Travel award worth CA$1,000.
  • Awarded to only about 11-20 students across Canada (domestic and international) in a calendar year.
  • Undertook two graduate-level courses in bioinformatics at the BC Cancer Agency.
  • Undergoing a research assistantship at Vancouver General Hospital to investigate relationship between imaging biomarkers in brain MRI scans and genetic findings in patients diagnosed with high or low grade gliomas.

NSERC-CREATE Topup

Faculty of Applied Sciences at SFU Sep 2018 – Sep 2018
  • One time award by the Faculty of Applied Sciences (FAS) at SFU worth CA$6,000 for NSERC-CREATE scholarship awardees.
  • Award covered partially by SFU and partially through RA by senior supervisor.

MSc Full Scholarship

School of Computing Science at SFU Sep 2017 – Apr 2019
  • Admitted to SFU’s Computing Science Graduate Program with guaranteed full scholarship for two years of study worth CA$33,000.

SFU Entrance Scholarship

School of Computing Science at SFU Sep 2017 – Sep 2017
  • Scholarship worth CA$3,000 awarded for excellent academic record, in conjunction with the full scholarship offer to SFU for CS graduate program.

Mentoring of Engineering Students by INAE Fellows

Indian National Academy of Engineering May 2016 – May 2016
  • Scholarship worth INR20,700 for undertaking bachelor’s dissertation under Prof. Sushmita Mitra at ISI Kolkata.
  • Awarded to only 47 students nationwide (India).

Summer Research Fellowship

National Institute of Technology Silchar May 2014 – Jun 2014
  • Scholarship worth INR6,000 for undertaking a research internship at NIT Silchar during sophomore year of undergrad, under Dr. Jayasree Chakraborty and Dr. Abhishek Midya.

Projects

*
All Deep Learning Reinforcement Learning Computer Vision Medical Imaging NLP

Comparison of Deep Reinforcement Learning Algorithms for Learning Control Policies for Physics-based Locomotion Tasks

Comparison of various state-of-art reinforcement learning algorithms

Evaluating Machine Translation Systems using Weighted Vote of Different Scoring Metrics

We propose a novel multi-metric voting scheme for evaluating machine translation systems.

Modelling Chinese Word Segmentation as Sequence to Sequence Prediction Problem

Re-imagining the problem of word segmentation as a sequence prediction problem, predicting spaces between words in a sequence.

Predicting Survival Time for Patients Diagnosed with Gliomas Using Compressed Representation of Brain MRI Scans

Convolutional neural networks based autoencoders for generating low-dimensional representations of brain MRI scans.

Automatic Characterization of Breast Masses using NSGA-II based Feature Selection

Propose a new feature extraction and selection method using NSGA-II for classifying masses as benign or malignant.

Recent & Upcoming Talks

Invited to present my research in which we proposed a multi-input, multi-output generative adversarial network (GAN) called MM-GAN as a …
May 20, 2019 Djavad Mowafaghian Centre for Brain Health, UBC
Poster Slides Video

Invited to present my research in which we proposed a multi-input, multi-output generative adversarial network (GAN) called MM-GAN.
May 1, 2019 3:30 PM Djavad Mowafaghian Centre for Brain Health, UBC
Slides

Presented an interesting paper from ICLR 2019 that performs out-of-distribution sample detection using a simple framework.
Apr 25, 2019 3:30 PM Burnaby, BC
Slides

Presenting a highly scientific paper in a way that high school students can understand. The paper is a state-of-art method that uses …
Nov 20, 2018 3:00 PM BC Cancer Agency, Vancouver, BC Canada
Slides

This talk covers a classical method for shape registration using approaches from information theory and deformations.
Oct 31, 2018 3:00 PM Burnaby, BC Canada
Slides

Paper presentation for the Live-Wire segmentation method, that was famously used in Adobe Photoshop.
Oct 10, 2018 3:00 PM Burnaby, BC Canada
Slides

Literature survey on the uses of text mining and natural language processing methods in bioinformatics.
Sep 20, 2018 3:00 PM BC Cancer Agency, Vancouver, BC Canada
Slides

Presented a state-of-art long-read assembly pipeline ‘Canu’, which builds upon many existing methods.
Aug 20, 2018 3:00 PM BC Cancer Agency, Vancouver, BC Canada
Slides

A talk regarding the current state-of-art in handling missing inputs for segmentation or classification problems.
Jul 12, 2018 1:00 PM Burnaby, BC Canada
Slides

Presented two seminal models of physics-based walking, called SIMBICON and GENBICON. The course instructor KangKang Yin is the inventor …
Mar 20, 2018 3:00 PM Burnaby, BC Canada
Slides

A mathematical, in-depth tutorial about reinforcement learning presented to the lab members. This was to facilitate members to take up …
Mar 14, 2018 3:00 PM Burnaby, BC Canada
Slides

This talk covers the first known application of reinforcement learning agents for landmark detection in CT Scans in medical imaging.
Mar 12, 2018 3:00 PM Burnaby, BC Canada
Slides

Featured Publications

MM-GAN

Magnetic resonance imaging (MRI) is being increasingly utilized to assess, diagnose, and plan treatment for a variety of diseases. The ability to visualize tissue in varied contrasts in the form of MR pulse sequences in a single scan provides valuable insights to physicians, as well as enabling automated systems performing downstream analysis. However many issues like prohibitive scan time, image corruption, different acquisition protocols, or allergies to certain contrast materials may hinder the process of acquiring multiple sequences for a patient. This poses challenges to both physicians and automated systems since complementary information provided by the missing sequences is lost. In this paper, we propose a variant of generative adversarial network (GAN) capable of leveraging redundant information contained within multiple available sequences in order to generate one or more missing sequences for a patient scan. The proposed network is designed as a multi-input, multi-output network which combines information from all the available pulse sequences, implicitly infers which sequences are missing, and synthesizes the missing ones in a single forward pass. We demonstrate and validate our method on two brain MRI datasets each with four sequences, and show the applicability of the proposed method in simultaneously synthesizing all missing sequences in any possible scenario where either one, two, or three of the four sequences may be missing. We compare our approach with competing unimodal and multi-modal methods, and show that we outperform both quantitatively and qualitatively.
PDF Poster Slides arXiv IEEE Xplore
ERGO

Single molecule localization microscopy (SMLM) allows unprecedented insight into the three-dimensional organization of proteins at the nanometer scale. The combination of minimal invasive cell imaging with high resolution positions SMLM at the forefront of scientific discovery in cancer, infectious, and degenerative diseases. By stochastic temporal and spatial separation of light emissions from fluorescent labelled proteins, SMLM is capable of nanometer scale reconstruction of cellular structures. Precise localization of proteins in 3D astigmatic SMLM is dependent on parameter sensitive preprocessing steps to select regions of interest. With SMLM acquisition highly variable over time, it is non-trivial to find an optimal static parameter configuration. The high emitter density required for reconstruction of complex protein structures can compromise accuracy and introduce artifacts. To address these problems, we introduce two modular auto-tuning pre-processing methods: adaptive signal detection and learned recurrent signal density estimation that can leverage the information stored in the sequence of frames that compose the SMLM acquisition process. We show empirically that our contributions improve accuracy, precision and recall with respect to the state of the art. Both modules auto-tune their hyper-parameters to reduce the parameter space for practitioners, improve robustness and reproducibility, and are validated on a reference in silico dataset. Adaptive signal detection and density prediction can offer a practitioner, in addition to informed localization, a tool to tune acquisition parameters ensuring improved reconstruction of the underlying protein complex. We illustrate the challenges faced by practitioners in applying SMLM algorithms on real world data markedly different from the data used in development and show how ERGO can be run on new datasets without retraining while motivating the need for robust transfer learning in SMLM.
IEEE Xplore

Recent Publications

Magnetic resonance imaging (MRI) is being increasingly utilized to assess, diagnose, and plan treatment for a variety of diseases. The …
PDF Poster Slides arXiv IEEE Xplore

Single molecule localization microscopy (SMLM) allows unprecedented insight into the three-dimensional organization of proteins at the …
IEEE Xplore

Inspired by the success of Convolutional Neural Networks (CNN), we develop a novel Computer Aided Detection (CADe) system using CNN for …
PDF arXiv

Skip connections in deep networks have improved both segmentation and classification performance by facilitating the training of deeper …
PDF arXiv

In this paper, a novel approach for the verification of offline handwritten signatures is proposed. Despite tremendous growth of …
PDF IEEE Xplore
See all publications

Popular Topics

artificial neural networks bioinformatics blstm classification computer animation computer vision convolutional neural network convolutional neural networks deep learning document analysis generative adversarial networks image processing invited linear algebra long-read assembly long short term memory networks lstm machine learning machine translation medical image analysis medical imaging mri natural language processing ngs-sequencing optimization physics-based control physics-based locomotion pid controller reinforcement learning sequencing shape registration spatial filtering super resolution microscopy svm