Work Experience

 
 
 
 
 
August 2022 – Present
Bellevue, USA

Software Engineer

Amazon Web Services (AWS)

Product: AWS Management Console

Team: User Insights Services Personalization

• Led a team of 4 engineers to deliver Machine Learning projects that generate real-time recommend services and user persona for AWS users. Unified the data source for consumption by all AWS Console teams who saw a 20% jump in new user activities in the Beta rollout.
• Designed a unified data service across all AWS teams, emitting data for personalized experience on different consoles. Captured data from RedShift into S3 using Hammerstone SQL jobs and built a data pipeline with Python Glue to process the data. Ran a SageMaker ML inference model via Java AWS Lambda and used DynamoDB as the model inference storehouse
• Added a data backfill workflow in Java for data sanity and collected metrics on Cloudwatch dashboards to trigger alarms for polluted data
• Opt-in Region Expansion: Working on expanding the Personalization service in multiple regions. Reduces service latency, mitigates data security concerns and provides a one-click mechanism to expand the service to any new AWS region in the future.
• Participated in code reviews and design discussions. Handled on-call duties for personalization and data engineering teams.


Technologies: Java, DynamoDB, Lambda, Simple Storage Service (S3), Simple Queue Service (SQS), Glue, Simple Notification Service (SNS), SageMaker, Identity & Access Management (IAM), CloudWatch, OpenSearch

Product: OpenSearch

Team: OpenSearch Core Engine

• Enhanced performance of AWS OpenSearch, an open source version of ElasticSearch in Java. Allowed deletion of read-only searchable snapshot indices and guarded snapshot backing indices from deletion. Fixed the search API for datetime input parameter.


Technologies: Java

 
 
 
 
 
March 2021 – August 2022
Redmond, USA

Software Engineer

Microsoft

Product: Whiteboard

Team: Whiteboard Integrations

• Worked on several high impact features that enhance Whiteboard user experience in Microsoft Teams having 270M+ monthly active users.
• Designed and implemented owner-only mode in React and Typescript. Extended Whiteboard edit mode to external users in tenant meeting.
• Improved Whiteboard security by introducing proof tokens on the client side and validating user actions in a Teams meeting session.
• Mitigated privacy concerns for gov tenants for Tab App by migrating data storage to the latest low-latency Microsoft Fluid framework.
• Mentored interns twice to enhance Whiteboard functionalities within Teams. The projects accomplished integrating the app to unify user-owned boards, pasting board links in meeting chat and implementing the Whiteboard notification feature.
• Actively participated in code reviews, design discussions, interview loops and mentoring new hires. Promoted writing high quality code with documentation and unit tests within the org. Increased diversity and inclusion awareness at Whiteboard as a member of D&I team.


Technologies: React, Typescript, C#

 
 
 
 
 
May 2020 – August 2020
Redmond, USA

Software Engineering Intern

Microsoft

Product: Whiteboard

Team: Whiteboard Integrations

Created a Whiteboard tab application for Microsoft Teams using C#, Typescript and Azure CosmosDB. This application led to ~10% increase in Microsoft Whiteboard usage at schools and organizations for asynchronous collaboration during Covid-19.


Technologies: C#, Typescript

 
 
 
 
 
October 2018 – July 2019
Hyderabad, India

Software Engineer

Paysafe

Product: Unity, a single API platform for the US iGaming and Gambling transactions

Team: Unity Portal

Contributions:

Merchant Management: Conceptualised and built an entire user management workflow for the Unity Business Portal that included stages starting from first time login, change password to forgot password for new merchants.

Merchant Onboarding: Improved the existing access management workflow by providing the functionality to assign pre-defined and custom roles and permissions for a hassle-free merchant onboarding.


Technologies: Java, AngularJS, Maven, MySQl, REST, Micro-services, Kafka, MapR, Highcharts

 
 
 
 
 
July 2016 – September 2018
Hyderabad, India

Software Engineer

D.E. Shaw

Team: Human Resources Development

Contributions:

Application Tracking System: Rebuilt the internal Application Management System (AMS) and Recruitment Management System (RMS) with 50% decrease in page load time and functionalities that improved the applicant experience and increased the efficiency of the Human Capital team.

Automated various internal processes such as Deferred Compensation management, Payroll Reconciliation and Employee Contribution by building user interactive workflows and gadgets. Worked on solving high-priority production issues with extremely low turn-around time.


Technologies: Java, MySQL, ReactJS, Javascript, Slickgrids, MyBatis

Team: Payroll Development

Contributions:

Automated various internal processes such as Deferred Compensation management, Payroll Reconciliation and Employee Contribution by building user interactive workflows and gadgets. Worked on solving high-priority production issues with extremely low turn-around time.


Technologies: Java, MySQL, ReactJS, Javascript, Slickgrids, MyBatis

 
 
 
 
 
December 2015 – May 2016
Pune, India

Research Project Intern

AlgoAnalytics

Brainoread: An open-source classifier to classify patients suffering from Schizophrenia and Dementia

Implemented an open-source classifier in R to classify patients suffering from brain disorders such as Schizophrenia and Dementia. Combined Caret (R), Weka (Java) and Scikit Learn (Python) in one package to achieve an accuracy of 92% (Schizophrenia, 119,748 test cases, 2nd best in Kaggle) and 98% (Dementia, 336 total cases). Used SVM, Neural Networks and many other algorithms ensembled into one after feature selection.


Technologies: R, Java, Python, Machine Learning

 
 
 
 
 
May 2015 – July 2015
Pune, India

Software Developer Intern

ADP

Contributions:

Built a Hiring Manager web application for tracking open positions at the firm as well as various candidacy stages like Resume Review, Resume Screening, Interviewed, in the form of dashboards and intuitive pie charts. Received a special mention from the Director of Global Product and Technology at the firm for the application.


Technologies: Java, AngularJS, HighCharts

Education

  • Master of Science in Computer Science (2020)

    Stony Brook University, Stony Brook, NY, USA

    CGPA: 3.79/4.0



List of courses taken

Fall 2019

  • CSE 548 - Analysis of Algorithms
  • CSE 519 - Data Science Fundamentals
  • CSE 518 - Human Computer Interaction
  • CSE 590 - Introduction to Modern Cryptography

Spring 2020

  • CSE 519 - Probability & Statistics For Data Scientists
  • AMS 559 - Smart energy in the information age
  • CSE 523 - Advanced Project in Computer Science
  • Bachelor of Technology in Computer Engineering (2016)

    College of Engineering Pune, Pune, India

    CGPA: 8.44/10.0

List of relevant courses taken

  • CT 205 - Data Structures
  • CT 09010 - Operating Systems
  • CT 14010 - Graph Theory
  • CT 09012 - Software Engineering
  • CT 09003 - Database Management System
  • CT 14002 - Storage and Virtualisation
  • CT 09004 - System Programming
  • CT 09005 - Computer Networks
  • CT 14001 - Compiler Construction
  • CT 208 - Data Communication
  • CT 204 - Microprocessor Techniques

Publication

Generic binary classifier tool for diagnosis of patients suffering from brain disorders in R

View Publication at IEEE Explore

Academic Projects

.js-id-mitesh

Missing Data Imputation using Generative Adversarial Networks (ongoing)


  • Building a Generative Adversarial Net (GAN) that accurately imputes missing data in categorical as well as numerical datasets as a part of my Masters’ advanced research project.
  • Achieved an RMSE of 0.052 on UCI spam base dataset and 0.126 on UCI letter recognition dataset.
  • Informative visualisations to get an overview of the success and errors of the imputor and for further improvement.
  • Currently using only a single layer of Generator and Descriminator, plan on introducing stacked ensemble.
Research Project | Team Size: 2 | Feb’20 - Present | Github Repo
Technologies: Python, Tensorflow, Machine Learning, Visualisation

Energy Disaggregation and Transfer Learning using Deep Learning


  • Built a neural network architecture that disaggregates individual appliance’s electricity consumption from the mains data using stacked ensemble learning.
  • Built using Bi-Directional LSTM and Bi-Directional GRU at its core.
  • Model trained on one appliance can also be used to segregate the consumption of another appliance from mains data using transfer learning. Eg. trained on Fridge, tested on kettle.
Smart Energy Course Project | Team Size: 3 | Mar’20 - May’20 | Github Repo
Technologies: Python, Keras, Deep Learning

COVID-19 Data Analysis


  • Proposed and tested 5 hypotheses (Eg. correlation between number of deaths across weeks) using Z-test, Chi-square, KS, Wald’s & Permutation Tests and Bayesian Inference on the COVID-19 NY counties data for the month of April.
  • Estimated county-wise deaths through time series prediction using EWMA & Auto Regression.
  • Also analysed the correlation between number of deaths and number of road accidents in the counties.
Prob Stats Course Project | Team Size: 6 | Apr’20 - May’20 | Github Repo
Technologies: Python

Retail Sales Data Analysis


  • Performed an in-depth analysis on retail sales data consisting of 17M rows and 60 features from Costello’s Ace, a hardware firm in the USA.
  • Provided valuable suggestions to the firm for increasing sales through time series analysis, market basket analysis and personalized product recommendations for customers.
  • Helped in business expansion by suggesting potential locations to open new stores. Backed up all findings through interactive visualisations using Tableau and Matplotlib.
Data Science Course Project | Team Size: 3 | Sep’19 -Dec’19
Technologies: Python, Tableau, Jupyter Notebook

Keyboard Swipe Recognition


  • Implemented the SHARK2 algorithm to correctly detect the word typed on the keyboard through a swipe.
  • Can currently detect a swipe for one of the 10,000 English words and can be extended easily to detect more words.
  • Detection takes very less time along with suggesting more similar pattern words.
Human Computer Interaction | Team Size: 1 | Sep’19 -Dec’19 | Github Repo
Technologies: Python, Flask | Live Demo

Brainoread


  • Implemented an open-source classifier in R to classify patients suffering from brain disorders such as Schizophrenia and Dementia.
  • Achieved an accuracy of 92% (Schizophrenia, 119,748 test cases, 2nd best in Kaggle) and 98% (Dementia, 336 total cases) using this binary classification tool by using SVM, Neural Networks and many other algorithms ensembled into one after feature selection.
  • Combined Caret (R), Weka (Java) and Scikit Learn (Python) in one package to efficiently use the models according to the preference.
  • Tuned classifier hyperparameters to increase classification accuracy from 85% to 92% while managing speed-accuracy trade-offs.
Research Project | Team Size: 3 | Dec’15 -May’16 | Github Repo
Technologies: R, Java, Python, Machine Learning

Skills

Java : 95%
Layer 1
C++ : 90%
Layer 1
C : 90%
Layer 1
Python : 85%
Layer 1
R : 80%
Layer 1
SQL : 95%
Layer 1
HTML/CSS : 85%
Layer 1
Javascript : 85%
Layer 1
Typescript : 90%
Layer 1
C# : 80%
Layer 1

Maven, Spring, Struts & RESTful Web Services95%

Native AWS80%

ReactJS & AngularJS80%

Data Science90%

System Design85%

Git & Jira95%

Swagger80%

Machine Learning85%