Data Science



Cross-Domain Context Prediction

Cross Domain Context Prediction for Sketch-Based Image Retrieval

SUMO Challenge

Facebook’s Scene Understanding and Modeling Challenge.

Betweenness Centrality for Streaming Graphs

Computing Betweenness Centrality on Streaming Graphs



Head Teaching Assistant

Georgia Institute of Technology

Aug 2019 – Present Atlanta, GA

Serving as Head TA for CS3600 - Intro. to Artificial Intelligence.

Responsibilities include:

  • Managing a team of 17 Teaching Assistants in grading coursework and proctoring exams for 450 undergraduate students
  • Developed entirely new mass autograder system with built-in plagiarism detector to reduce the time each TA must spend grading from 2 hours to 15 minutes.
  • Served as guest lecturer on Particle Filtering for a class section with 300 students
  • Creating new material with detaile dexplanations of concepts such as Admissible & Consistent Heuristics and Mutual Information and with additional examples to ensure students have a firm grasp of course material
  • Leading review sessions to help students better understand course concepts in preparation for projects and exams

Computer Vision Intern

The Home Depot

May 2019 – Aug 2019 Atlanta, GA
  • Fine-tuned 20+ models to extract fine-grained features from furniture images to enable training of a generative model that could produce aesthetically compatible furniture collections
  • Developed a reusable framework for quickly aggregating images from the various data sources, performing data cleanup and augmentations, fine-tuning ResNet/VGG models, and finally evaluating using F1 scores and confusion matrices


Duke TIP

Jun 2018 – Aug 2018 Houston, TX & Atlanta, GA
  • Taught a total of 40 students for four weeks across two courses: Modern Programming and Artificial Intelligence
  • Designed syllabi, conducted parent-teacher conferences, crafted assignments spanning five core topics: Fundamentals of Python Programming, Web Development, Pathfinding, Machine Learning, Natural Language Processing
  • Find out more about the courses I created here

Software Engineer


Jun 2017 – Jun 2018 Atlanta, GA
  • Automated a data collection task that consumes 90 days’ worth of data for researchers to create adaptive fraud detection models
  • Optimized memory consumption for a new service to use 75% less memory than previous implementations
  • Created new framework for continuous deployment of fraud detection models to production systems
  • Migrated database records from Postgres to Redis and used custom key-value encoding to optimize lookup time

Recent Publications

Quickly discover relevant content by filtering publications.

Recent Posts

In Fall 2018, I decided to apply to be a speaker for TEDxGeorgiaTech's Fall speaker salon. I was chosen as one of the 6 speakers among …