Vaibhav Agrawal

Software Engineer

Meta, Bellevue WA

Building distributed ad measurement systems for Facebook and Instagram

Software Engineer

Twitch, Seattle WA

Subject Matter Expert for Twitch's distributed payments system that orchestrates "X-million $" transactions daily

Software Developer (Part time)

FITL, New York University

Built a RESTful API driven web app (Django, React.js, SQL) to manage the undergraduate tutoring program with a dedicated algorithm for matching students and tutors into individual and shared tutoring sessions

Associate Product Manager Intern

Led cross-functional effort to launch a centralized tool to consolidate fragmented release schedules

Summer Research Intern (Statistics)

Indian Institute of Science, Bangalore

Analyzed Google's traffic speed data to explore statistical distributions of link-level traffic speed on transportation networks and developed a dedicated algorithm for the task of link extraction leveraging from Google Maps API

Indian Institute of Technology Kharagpur

GPA: 9.03/10 | Gold Medalist

Algorithms and Data Structures, Probability and Statistics, Switching Circuits and Logic Design, Computer Architecture and Operating System, Machine Learning, Artificial Intelligence, Natural Language Processing, Parallel Programming

Crash Severity Analysis Through Nonparametric Machine Learning Methods

Published in the Journal of the Eastern Asia Society for Transportation Studies, 2019 Volume 13 Pages 2614-2629.

Keywords: Crash severity, Data Mining, Traffic Crashes, XGBoost; Cited by 9 [PDF]

Improving the learnability of classifiers in Sanskrit OCR corrections

Published in the Computational Sanskrit & Digital Humanities, 2018 Pages 143-161.

Presented at the 17th World Sanskrit Conference at Vancouver Canada in July 2018.

Keywords: OCR, Deep Learning, NLP, Attention Models; Cited by 7 [PDF]

Twitter Bot Detection

Built a machine learning model on a big data scale (accuracy of 91%) on Spark ML, MongoDB and Kafka to identify Twitter bots that spread political propaganda. [Github]

Intelligent System Design using Cloud

Email Spam Detector is a serverless, microservice driven application, powered by a machine learning model to predict whether an email is spam or not. This system that upon receipt of an email message, automatically flag it as spam or not, based on the prediction obtained from the ML model. And it replies back to the sender of the email with a reasonable prediction. [Github]

NLP Powered Photo Album Search

Voice Based Photo Album Search is a serverless, microservice driven web-based application, that can be searched using natural language through both text and voice. We can upload photos and search then similar to a image search engine. [Github]

SMART DOOR AUTHENTICATION SYSTEM

Smart-Door-Authentication-System is a serverless, microservice driven web-based application. It builds a distributed system that authenticates people and provides them with access to a virtual door. [Github]

Dining Concierge Chatbot

Dining Concierge Chatbot is a serverless, microservice driven web-based application. It is an intelligent natural language powered chat-bot that can help you provide restaurant suggestions based on your requirements such as - City, Time, Number of people, Cuisine Type and Date.

Database Designing

Designed the schema and created a normalized apartments database with 5 tables linked together using keys. [Github]

Unix Shell

Implemented my own Shell from scratch in XV6 operating system using UNIX system calls as a part of Operating Systems course at New York University. Command execution, I/O Redirection and Pipelining. [Github]

Lottery Scheduler

Implemented and Tested lottery scheduling, a randomized algorithm that allows processes to receive a proportional share of the CPU without explicitly tracking how long each process has been run. [Github]

Automated clustering of Research Papers

AAAI wants an automated unsupervised script to group research documents into different clusters, so that all papers having similar high-level domains will be grouped together. [Github]

Stance detection from tweets

The project aims at identifying different stances adopted by the people, concerning various aspects of cancer (such as lethality, prevention, causes, etc) from an extensive collection of tweets. [Github]

Post OCR Processing

Designed an algorithm for word corrections of post OCR Sanskrit scripts through statistical language models using edit distance on a known dictionary. Improved the accuracy of the algorithm up by 10%, by introducing morphological language constraints from the grammar, using Sandhi. [Github]

Android Apps

Worked on number of projects while completing the Android Development Course on Udacity placed by Google. Developed variety of apps that uses basic android concepts of user interface and user inputs. [Github]