SUMEET RANKA

Hello. Thanks for stopping by.

Currently a Data & Applied Scientist at Microsoft Canada, Vancouver.
My interests involve Applied Machine Learning and Software Development. I have experience working with various modalities such as images, texts and clinical datasets.

I completed my Masters degree from the University of Toronto in the Department of Computer Science in Applied Computing and my Bachelors degree from IIT Guwahati with a major in Computer Science and Engineering. Previously, I have worked in Huawei (Noah Arks Lab) as an Associate Researcher and at Samsung Research Institute as a Research Engineer.

When I am not coding I am mostly exploring new places, dancing or reading.

Here is my CV. You can contact me through any of the medium mentioned below. Have Fun!


Experience

Data & Applied Scientist 2

Bing Multimedia, Microsoft Canada, Vancouver

Contributing to a new product, PRISM Feed. It will be the default experience on Bing Image Search

  • Owner of module for diversifying and re-ranking recommended documents for a new product called PRISM
  • Developed scalable, modular code from the scratch. Shipped to en-* market
  • Working on Query Classifier for identifying relevant user signals for personalization
Tools: C#, PowerBI, Internal Compute and Development Tools

Jul 2021 - Present

Associate Researcher

Huawei Technologies Canada, Toronto

Developing machine learning models trained over images and videos and then deploying them on resource-constraint devices.

  • Prototyped and deployed face detection and object detection models on Huawei NPUs
  • Mentored four co-op intern students working on different ML inference projects
  • Contributor to drowsiness detection system for cars, featured as a major product by Huawei CBG CEO
  • Consulted on multiple projects for architectural design and embedded systems related optimization
  • Tools: C++, Python, Tensorflow, Android JNI

Feb 2020 - Jul 2021

Applied Research Intern

Phenomic AI

The internship aimed at analyzing the microscopy images of multi-cell cultures and extract intensity, shape, and texture-based features using machine learning techniques to quantify the cell-cell interactions.

  • Developed and deployed a robust and scalable segmentation and feature extraction pipeline for microscopic images.
  • Classified and analysed different cells identified through segmentation using supervised and semi-supervised techniques.
  • Quantified and validated the contact-dependent effects between lung cancer cells and fibroblasts.
Tools: Python, AWS services, CellProfiler, ImageJ (Fiji)
Links: Poster

May 2019 - Dec 2019

Research Engineer

Samsung Research Institute - Bangalore

Sensor-based In-house Tracking System

  • Devised an offline tracking method for residents based on sensors in the house
  • Published in a Springer conference
  • Developed a robust generic Scala script to preprocess raw sensor-based data. Preprocessing time reduced from two weeks to three days
Tools: Apache Spark, Scala, Python, NumPy
Award: SPOT Award for the preprocessing script

Voice Activity Detection

  • Literature review on the current state-of-the-art human presence detection methods
  • Implementation of the feature extraction to optimize in memory and time
  • Used TensorFlow Lite to validate the output of the human presence model on a Tizen-based smart speaker
Tools: C++, Python, TensorFlow, TensorFlow Lite
Award: SPOT Award for the reduction of the inference time from 7 sec to 2.9 sec

Jun 2017 - Aug 2018

Software Engineering Intern

Flipkart Pvt. Ltd.

  • Developed functional and performance testing modules for a hierarchical database architecture
  • Identified bottlenecks and limitations in the underlying database architecture
  • Modularised to configure and develop new testing modules easily
Tools: Apache JMeter

May 2016 - Jul 2016

projects

k-space Imputation and MRI Reconstruction

Dr. Ben Fine, Trillium Health Partners; Prof. Marzyeh Ghassemi, University of Toronto

Explored denoising autoencoder based U-Net and perceptual GAN to impute k-space to improve the process of MRI reconstruction. Proposed the DAE-UNet method and trained using fastMRI dataset.

  • Idea is to impute k-space first and then reconstruct image.
  • Observed that as the area of interest in a MRI image is much less as compared to the size of the entire image, it might be possible that a model learns to reconstruct the background and still get a low MSE
Team: Vaibhav Saxena, Pulkit Mathur, Duc Truong

Feb 2019 - Apr 2019

The Tempest

Prof. Karan Singh, University of Toronto

Theatrical presentation of the play, The Tempest, on a 180-degree curved screen leveraging the concept of proprioception and 3-D drawing for the character Ariel

  • Inclusion of proprioception to trigger the events based on zones selected instead of explicit display of menu items
  • Performed comparative analysis between pre-defined and user-defined zones to understand the user’s comfort level
Team: Andrejs Ru, Sahil Narula, Sneha Desai
Tools: Unity, C#

Sep 2018 - Dec 2018

Text Readability Analysis using Language Models

Prof. Ashish Anand, IIT Guwahati

  • Designed a new fine-grained, computational measure of readability. Main conjecture was that the predictability of a text, as determined by standard language models, is a viable metric of its readability
  • Evaluated our metric on various datasets with known readability scores, such as simple and standard Wikipedia articles, labeled datasets from prior work on automatic readability scoring systems, and even created a new dataset by parsing freely available school textbooks
  • Obtained good results for all the three datasets, unlike earlier approaches
Tools: Python

Feb 2017 - Apr 2017

Using Spatial Transformer Networks for Egocentric Images

Dr. Arijit Sur, IIT Guwahati

  • Implemented spatial transformer networks for object recognition from egocentric images
  • Evaluated it on GTEA (Georgia Tech Egocentric Activities) dataset. The model showed better results than a traditional CNN model
Tools: Python, TensorFlow

Sep 2016 - Nov 2016

Assistive Device for Hand Amputees

Dr. S. B. Nair, IIT Guwahati

ADHA is a device that is designed to support people who have had both their arms amputated above or below the elbow. It provides two major types of support:

  • Enable the user to connect to the digital world by forming a convenient interface to a computer's keyboard and mouse.
  • Enable the user to transport heavy objects by having a transport robot that they can control.
Tools: Arduino, Python

Oct 2015 - Nov 2015

Battle of Hogwarts

Pet Project

A Harry-Potter themed MUD (Multi-User Dungeon) game based on Evennia Platform (a python-based development system). Developed as part of the Technothlon event, organised by Techniche (IIT Guwahati) for senior-secondary school students.
PS: It was fun to write an extended version of Harry Potter series for this game.


Skills

Languages

  • C
  • C++
  • Python
  • Scala
  • C#

Big Data Tools/Libraries

  • TensorFlow
  • NumPy
  • SciPy
  • spaCy
  • Apache Spark

Frameworks/IDE

  • Visual Studio
  • Unity
  • Django
  • JupyterLab
  • Android JNI

Miscellaneous

  • LaTeX
  • Git
  • PowerBI
  • Docker
  • MySQL
  • CellProfiler
  • ImageJ (Fiji)


Publications

USHEr: User Separation in Home Environment

Sumeet Ranka, Vishal Singh, Mainak Choudhury
Published at: International Conference on Smart Homes and Health Telematics (ICOST), 2018

Abstract:
With the increase in presence of smart devices in our daily life, it is an important problem for these devices to be more intelligent. The most sought after problems in this area are activity recommendation and prediction. Researchers have proposed solutions for this problem, however, most of them are based on single-user home space. In this paper, we propose an unsupervised approach to separate the logs of multi-user home space into buckets equal to the number of users. With a minimal set of assumptions, the aim of the method is to transform the multi-user problem to a single-user problem. It is achieved by estimating the layout of the house and then tracking the users at room-level. We achieved empirically-determined high precision in estimating the layout and 74% accuracy in separating the multi-user stream.

Links: [paper]


Talks

Making Homes Smarter using Machine Learning

Delivered as a guest speaker in QIP-STC 2017, conducted by Indian Institute of Technology (BHU) Varanasi, titled on Machine Learning: Trends, Perspectives, and Prospects

Slides cannot be made public

Campus Network Architecture

Tech-Talk in Indian Institute of Technology Guwahati explaining and answering questions about the intricate details of the internal network of the college. It was an interest-based exploration that resulted in a talk.

Links: Slides

Object Oriented Programming

Another tech-talk out of interest for the final year undergraduate and graduate students on the basics of Object Oriented Programming.


Contact

email: firstname [dot] lastname [dot] 47 [at] gmail [dot] com

address: Vancouver, Canada