University of Toronto

Master in Science In Applied Computing (Data Science Concentration) Sept 2018 - present

Courses Taken :

  • Machine Learning and Data Mining (CSC411, Fall-2018)
  • Deep Reinforcement Learning (CSC2541, Fall-2018)
  • Data Science Methods, Collaboration, and Communication (STA2453, Fall-2018)
  • BlockChain (CSC2125, Fall-2018)
  • Natural Language Computing (CSC2511, Winter-2019)
  • Machine Learning In Health Care (CSC2541, Winter-2019)
  • Thapar Institute of Engineering and Technology, Patiala

    Bachelor in Computer Science July 2013 - June 2017

    GPA: 9.0/10



    Software Engineer July 2017 - Aug 2018

  • Time Series Analysis - Predictive Analysis on a time series data dealing with programming and equipment issues reported so far. Thus, helping the firm to decide the requirement of maintenance staff and the cause of the issue. Link: github
  • Value Added Messaging - End to end pipeline from Mainframe till AWS Lambda for sending messages to the customer. We handled terabytes of data and used technology like Apache Kafka, Nifi, AWS ,EMR,EC2,SNS Link:

  • Achievements
  • Semifinalist in Code Grind (Hackathon) organized by HSBC at international level
  • BlackRock

    Software Engineer Intern Jan 2017 – Aug 2017

  • Automation tools- Developed a automation tool for testing portfolio management and other Java based software. Thereby replacing Licensed automation tool like UFT and ALM.
  • Achievements
  • Won Hack-Knight organized by BlackRock for 2017 interns- Created a web-based game application build using a game engine
  • Cleared Regional level and pitched our team idea at the Global level during Hackathon-2017 - Created a centralized web page to search for any query related to the firm.
  • Skills and Certificates

  • Certificate - R, Big Data, Data Analysis, Android, Cloud Technology from Infosys and HSBC (link)
  • Programming and Databases - Java, Python, R, C, C++, NodeJS, HTML/CSS, MySQL, MongoDb
  • Frameworks and Libraries - AWS, Hadoop, Tensorflow sklearn, nltk
  • Other Achievement - Two Bronze medals in international level coding competition @HACKERRANK (Profile)
  • Projects

    Evaluating Generative Adversarial Networks (GANs) using Effcient Real-World Estimate

  • A machine learning project where we used a MCMC sampling method using Metropolis-Hastings algorithm for obtaining samples from real-world density as learned by the discriminator of a GAN, using samples from the generator.
  • We improved upon an existing rejection sampling approach with respect to a higher acceptance percentage and less samples rejected around the modes.
  • Tools: Tensorflow
    Links: Documentation

    Voice Banking

    Using AWS and Alexa solved three use cases- Banking, Marketing and Product Description. Users can now inquire about their bank balance, search for nearby offers, clear credit card balance and can also do other banking tasks by just saying few words.
    Tools: NodeJS, AWS


    Adversarial Imitation Learning by Planning
  • A reinforcement learning project that proposes a framework to improve the state of the art imitation learning method (End-to-End Differentiable Generative Adversarial Imitation Learning).
  • It uses the concept of switching between model-based planning and model-free learning (DYNA) from expert data for better sample efciency.
  • Tools: Tensorflow
    Links: Documentation

    Deducto (Co-founder and Lead developer)

    Developed an Android app to provide marketing platform for local shop vendors for attracting customers based on the offers. Shop owners can now post their offers and customers can avail them by using the application.
    Tools: Java, Android
    Links: Link


    Design of 3-factor strict local language in ethological data mining of bengalese Finch's and White-Rumped Munia's song

    Pulkit Mathur , Vishwajeet Dagar , Ajay Kumar
    Published in: 2017 13th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery (ICNC-FSKD)
    The aim of the paper is to extract and compare the behavioral units of Bengalese Finch and White-Rumped Munia's songs and represent them by a scanner which is a counter automaton of sub-regular language. Scanner allows us to recognize the bird songs at a fast rate. We have used ethological data mining and automata-based approach for finding the rules governing the structure of bird songs. We have demonstrated the difference in the results obtained for both the birds. The results indicate that it is possible to represent these songs by using 3-factor strictly local language.
    Link: [paper]

    Client Testimonials

    • "Pulkit notably within 6 months provided back to back innovations on voice enabled banking products apart from his day job as a business data architect. interestingly while others tinker with technology and tools , Pulkit has an unique insight to find a business use case or solution to a problem with the innovation product he develops. something he called applied innovation. he has just within 2 weeks developed a pin enabled voice banking on nodejs that is simply brilliant . although he is decided to go for the academic route...he will be thoroughly missed and would be an asset to an team he works in. My best wishes"

      Nupur Mukerjee (Global head Hsbc ingestion partnership at Hsbc Global Technology Services Limited., Pune)
    • "Diligent and innovative" are the qualities that characterizes Pulkit in a genuine sense. He worked with me amid the time duration from January 2018 to July 2018. He always demonstrated commitment towards his work. Meeting deadlines, writing effective and industry standard code is something he always adheres to. He is not only a voracious learner who is always trying to grasp knowledge from domain experts, but also engages himself in the discussions on ways to enhance existing systems by bringing in latest technologies. He also exhibits leadership qualities which can be inferred from the fact that the team lead by him won the interns Hackathon contest. But the most important quality that sets him apart is the innovative ideas that he brings to the table; a live example of which is his POC regarding "Common search portal" that received applauds from the higher management and brought laurels to the team. Pulkit would be a true asset for any specialized group and I would truly recommend him to all organisations he would be part of in future.

      Vishal Jain, (Vice President at BlackRock)

    Pulkit Mathur. All rights reserved.