Karthik Bhaskar

Karthik Bhaskar

Machine Learning Researcher | Data Scientist | Software Engineer

University of Toronto

Vector Institute

Welcome!

I am a Senior Data Scientist at CIBC. Previously, I worked as a ML Researcher at WangLab affiliated with Vector Institute and University Health Network, proudly advised by Prof. Bo Wang. I completed my Master’s degree in ECE, specialized in Machine Learning at the University of Toronto advised by Prof. Deepa Kundur and by Prof. Yuri Lawryshyn. In my free time I work on projects at the intersection of Machine Learning, Natural Language Processing and Healthcare. Apart from my that, I am also interested in Deep Learning and Deep Reinforcement Learning with the focus of transfer learning, imitation learning, model-based RL. My ultimate goal is to build robust, privacy-preserved, and interpretable algorithms with human like ability to generalize in real-world environments by using data as its own supervision.

I am a Stream Owner and Discussion Group Lead of the “Machine Learning in Healthcare” stream at AISC (Aggregate Intellect), where we discuss one paper at a time every week from basics to state-of-the-art ML papers in HealthCare. I also host several live sessions with global researchers spanning the broad area of ML in HealthCare.

Throughout my life, I have approached every challenge with enthusiasm, creativity, and a ceaseless desire to achieve success. This passion and drive have paved the way to countless opportunities, unique experiences, and excellent relationships, both personally and professionally. I enjoy working with people and discussing ideas. If you would like to chat, feel free to send me a message on Twitter.

Outside of academics, I enjoy basketball, hiking, biking, running (pretty much every sport !).

Interests

  • Deep Learning
  • Privacy Preserved Trustworthy Machine Learning
  • Deep Reinforcement Learning
  • Natural Language Processing
  • Recommendation Systems

Education

  • M.A.Sc in Electrical and Computer Engineering/Machine Learning, 2020

    University of Toronto

  • B.Eng in Electronics and Communication Engineering, 2016

    Anna University

Experience

 
 
 
 
 

Data Scientist

CIBC

Nov 2020 – Apr 2022 Toronto
  • Built a Semantic Search Engine using Transformers to get relevant data for downstream ML tasks like building Dashboard.
  • Deployed the Search Engine using Streamlit, Docker, Docker Compose and other MLOps tools.
  • Built Transformer based Topic Modelling on Deficiency data and used the results in downstream Classification Model to predict the category of deficiencies.
  • Built a DistilGPT-2 based Transformer neural network to generate synthetic financial data for downstream ML models.
 
 
 
 
 

Machine Learning Researcher

Vector Institute

Aug 2020 – Jan 2021 Toronto
  • WangLab is affiliated with Vector Institute and University Health Network - Advised by Bo Wang
  • Worked on a project at the intersection of Computational Biology, Deep Learning and Natural Language Processing.
  • Applied Self Supervised Learning, Weak Supervision and Data Programming on MIMIC III database.
  • Built Transformer based architecture model to improve the accuracy of massively multi-task classification problem.
  • Our Team BeatCovid designed and built a DL model to predict the COVID-19 cases across worldwide in XPrize’s Pandemic Response Challenge and secured 16th place out of 250 teams participated in this challenge worldwide.
 
 
 
 
 

Machine Learning Engineer

Diversio

May 2020 – Aug 2020 Toronto
  • Builta text classifier using Word Embedding, LSTM, and Transformers.
  • Created a Diversity and Inclusion score that reflects employee experience within organizations.
  • Designed a Deep Learning based Recommendation System using the diversity and inclusion score.
 
 
 
 
 

Machine Learning and Data Science Intern

TD Canada Trust

May 2019 – Aug 2019 Toronto
  • Worked with cybersecurity professional and built a defense mechanism using deep learning and unsupervised learning techniques to prevent cyberattacks.
  • Built Unsupervised Auto Tagging algorithm and Automatic Rule Synthesis for Octavius.
  • Investigated an Automatic Rule synthesis algorithm for Octavius using Deep Reinforcement Learning to improve overall defense mechanism.
 
 
 
 
 

Graduate Machine Learning Researcher

University of Toronto

Aug 2018 – Jan 2020 Toronto

  • Responsible for developing and building cutting edge state of the art deep learning based recommendation system.

  • Built a Deep Learning-based Recommendation System for Wolseley’s e-commerce website from scratch to production.

  • Dataset is massive involving more than 200,000 unique customers and 500,000 unique SKUs

  • Achieved a personalized NDCG score of 72.4% and improved theO ne-Product Hit Ratio to 100%.

  • Deep Learning, Matrix Factorization, Collaborative Filtering, NLP, Bayesian Optimization, etc.

 
 
 
 
 

Machine Learning and Data Science Intern

TD Canada Trust

May 2018 – Aug 2018 Toronto
  • Worked with cybersecurity professionals and built machine learning-based classifiers to prevent cyberattacks.
  • Devised a Deep Learning model that detects and prevents the cyberattack before it happens (ProjectX).
  • Used NLP, Neural Networks, Knowlege Graphs for Keyword Extraction, etc.
 
 
 
 
 

Software Engineer

Infosys

May 2016 – Jul 2017 Chennai
  • Project deals with retail side applications supported by Apple open to outside as well as Internal to Apple.
  • Performed root cause analysis and carried out recommended changes.
  • Conducted impact analysis on new changes before moving the code to production systems.
  • Integrated changes in the production environment with minimal impact to existing functionalities.
  • Extensively monitored the system to make sure business users are not impacted.

Talks

Deep Learning in HealthCare and its Practical Limitations

This talk is about the promise of Deep Learning in Healthcare and its practial limitations. What are some of the gaps between academic research and prodcution level code and how can we mitigate this production level gap in Deep Learning and Healthcare, and what are some of the tools and techniques we can deploy to make production easier for Deep Learning especially Healthcare?
Jan 20, 2021 7:30 PM Virtual Talk
PDF Slides Video Follow
Deep Learning in HealthCare and its Practical Limitations

Blog

Thoughts on AI

Incremental Few-Shot Learning (IFSL)

What is Few Shot Learning? The human visual system exhibits the remarkable ability to effortlessly learn novel concepts from only one or a few examples and reliably recognize them later on.
Last updated on Jun 4, 2021 11 min read

Machine Learning Production, Machine Learning Research and Software Engineering

Machine Learning Production Vs Machine Learning Research Most people who are experts in Machine Learning gained their experience through academia: taking courses, doing research and reading research papers. Most companies nowadays use Machine Learning, although it is still relatively new compared to traditional software engineering.
Last updated on Jun 4, 2021 9 min read

Private AI - "Society Runs on Information Flows"

The main topic of the article is the privacy-transparency trade-off and how it affects a huge number of issues. This article walks through some of the most important challenges to society and identifies how the privacy-transparency trade-off underpins them.
Last updated on Jan 11, 2021 12 min read

Projects

.js-id-Reinforcement-Learning
Reinforcement Learning Deep Learning Machine Learning All

Amazon Fashion Recommender

Amazon Fashion Recommender Used Amazon Product Advertising API to gather data locally and used those files for Recommendation Used Two Approaches Natural Language Processing on Product Description Convolutional Neural Network on Product Images Look at the Notebooks for more details
View Code/Project
Amazon Fashion Recommender

Cancer Diagnosis

Personalized Medicine Redefining Cancer Treatment A lot has been said during the past several years about how precision medicine and, more concretely, how genetic testing is going to disrupt the way diseases like cancer are treated.
View Code/Project
Cancer Diagnosis

Collaboration and Competition: Multi Agent Deep Deterministic Policy Gradient

This repository contains my work for Udacity’s Deep Reinforcement Learning Nanodegree For this project, we will work with the Tennis environment. In this environment, two agents control rackets to bounce a ball over a net.
View Code/Project
Collaboration and Competition: Multi Agent Deep Deterministic Policy Gradient

Continuous Control: Deep Deterministic Policy Gradient

This project repository contains my work for the Udacity’s Deep Reinforcement Learning Nanodegree Project 2: Continuous Control. Project’s goal In this environment, a double-jointed arm can move to target locations. A reward of +0.
View Code/Project
Continuous Control: Deep Deterministic Policy Gradient

Defense GAN & Physical Adversarial Examples

Abstract This project discuss the transferability of state of the art defense techniques for adversarial examples for deep learning systems in the physical domain. The paper explores using adversarial attacks using the Fast Gradient Sign Method (FGSM), Carlini & Wagner (CW) and DeepFool attacks to generate adversarial images that are given to the classifier as a digital and physically transformed image.
View Code/Project
Defense GAN & Physical Adversarial Examples

Detecting Deforestation from Satellite Images

Project Description A Deep Learning approach to detecting deforestation risk, using satellite images and a deep learning model. We relied on Planet imagery from two Kaggle datasets (one from the Amazon rainforest and another on oil palm plantations in Borneo) and trained a ResNet model using FastAI.
View Code/Project
Detecting Deforestation from Satellite Images

Facebook Social Network Graph Prediction

Social network Graph Link Prediction - Facebook Challenge Problem statement: Given a directed social graph, have to predict missing links to recommend users (Link Prediction in graph) Data Overview Taken data from facebook’s recruting challenge on kaggle https://www.
View Code/Project
Facebook Social Network Graph Prediction

Google - Advertisement Click Prediction

Problem Description: Clickthrough rate (CTR) __ is a ratio showing how often people who see your ad end up clicking it. Clickthrough rate (CTR) can be used to gauge how well your keywords and ads are performing.
View Code/Project
Google - Advertisement Click Prediction

Human Activity Recognition

This project is to build a model that predicts the human activities such as Walking, Walking_Upstairs, Walking_Downstairs, Sitting, Standing or Laying. This dataset is collected from 30 persons(referred as subjects in this dataset), performing different activities with a smartphone to their waists.
View Code/Project
Human Activity Recognition

Human Like Chess Engine

Abstract We used supervised training to create a series of chess engines based on humans play at different levels of skill. We compared them to other engines and to human players and found that self-play trained engines would sometimes behave more human-like than the supervised ones, although we believe this may be due to improper hyperparameter selection.
View Code/Project
Human Like Chess Engine

Hyper Face

Abstract A multi-task learning convolutional neural network for the purpose of performing landmark localization and other correlated tasks is studied and analysed in this project. A different and more challenging task around landmark localization than the one implemented originally is studied using a HyperFace architecture.
View Code/Project
Hyper Face

Microsoft Malware Detection

Microsoft Malware Detection Problem Statement: In the past few years, the malware industry has grown very rapidly that, the syndicates invest heavily in technologies to evade traditional protection, forcing the anti-malware groups/communities to build more robust softwares to detect and terminate these attacks.
View Code/Project
Microsoft Malware Detection

Music Generation

Multi-layer recurrent neural networks for training and sampling from texts, inspired by karpathy/char-rnn. Requirements This code is written in Python 2, and it requires the Keras deep learning library. Usage All input data should be placed in the data/ directory.
View Code/Project
Music Generation

Navigation: Deep Q Networks

Deep Reinforcement Learning : Navigation This project repository contains my work for the Udacity’s Deep Reinforcement Learning Nanodegree Project 1: Navigation. Project’s goal In this project, the goal is to train an agent to navigate a virtual world and collect as many yellow bananas as possible while avoiding blue bananas
View Code/Project
Navigation: Deep Q Networks

Netflix Movie Recommendation System

Problem Description Netflix is all about connecting people to the movies they love. To help customers find those movies, they developed world-class movie recommendation system: CinematchSM. Its job is to predict whether someone will enjoy a movie based on how much they liked or disliked other movies.
View Code/Project
Netflix Movie Recommendation System

Quora Question Pair Similarity

Quora Question Pair Similarity Quora is a place to gain and share knowledge—about anything. It’s a platform to ask questions and connect with people who contribute unique insights and quality answers.
View Code/Project
Quora Question Pair Similarity

Self Driving Cars

Problem Definition We are here building a minimal version of self driving car. Here, we have a front camera view. This will transfer input to the computer. Then Deep Learning algorithm in computer predicts the steering angle to avoid all sorts of collisions.
View Code/Project
Self Driving Cars

StackOverflow Tag Predictor

Stack Overflow - Tag Prediction Description Stack Overflow is the largest, most trusted online community for developers to learn, share their programming knowledge, and build their careers. Stack Overflow is something which every programmer use one way or another.
View Code/Project
StackOverflow Tag Predictor

Taxi Demand Prediction

Taxi Demand Prediction Data Information Ge the data from : http://www.nyc.gov/html/tlc/html/about/trip_record_data.shtml (2016 data) The data used in the attached datasets were collected and provided to the NYC Taxi and Limousine Commission (TLC)
View Code/Project
Taxi Demand Prediction

Publications

October 2020

Implicit Feedback Deep Collaborative Filtering Product Recommendation System

In this paper, several Collaborative Filtering (CF) approaches with latent variable methods were studied using user-item interactions to capture important hidden variations of the sparse customer purchasing behaviours. The latent factors are used to generalize the purchasing pattern of the customers and to provide product recommendations. CF with Neural Collaborative Filtering(NCF) was shown to produce the highest Normalized Discounted Cumulative Gain (NDCG) performance on the real-world proprietary dataset provided by a large parts supply company. Different hyperparameters were tested using Bayesian Optimization (BO) for applicability in the CF framework. External data sources like click-data and metrics like Clickthrough Rate (CTR) were reviewed for potential extensions to the work presented. The work shown in this paper provides techniques the Company can use to provide product recommendations to enhance revenues, attract new customers, and gain advantages over competitors.

Preprint PDF

Skills

Actor Critic Methods CNN Collaborative Filtering Computer Vision CTR Cybersecurity Data FlyWheel Deep Deterministic Policy Gradient (DDPG) Deep Learning Deep Q Networks (DQN) Deployment Ensemble Methods Face Detection FastAI Feature Engineering GAN Google BigQuery Graph Networks Healthcare Hyperparameter Tuning Keras KL Divergence Landmark Localization LSTM Machine Learning MLOps Multi Agent Deep Deterministic Policy Gradient (MADDPG) NLP NLTK OpenCV Out of Domain Shift Analysis Policy Based Methods Python PyTorch Random Forest Recommendation System RecSys Reinforcement Learning ResNet RNN SQL StreamLit SVD SVM Tensorflow Unity ML Value Based Methods Visualization XGBoost

Contact

Drop a message!