cv

Basics

Name Kartik Sharma
Label Ph.D. student
Email ksartik@gatech.edu
Phone (470) 437-2059
Url https://ksartik.github.io/

Work

  • 2023.05 - 2023.08
    Research Intern
    Visa Research
    Developed a plug-and-play method to classify nodes using a personalized scope of its local neighborhood. Experiments show improved robustness, interpretability, and heterophily performance (submitted to ICML 2024).
    • Interpretability
    • Graph Neural Networks
  • 2018.09 - 2021.07
    Founding Member
    Tensor Dynamics Pvt. Ltd.
    Developed initial code-base for a green-tech startup to provide reliable solar energy forecasts. Forecasting models deployed by Load Dispatch Centre to monitor over 20 solar stations across India
    • Forecasting
    • Renewable Energy

Education

  • 2021.08 - Present

    Atlanta, GA

    Ph.D.
    Georgia Institute of Technology
    Reliability and controllability of ML models for structured data
  • 2017.07 - 2021.05

    New Delhi, India

    B.Tech.
    Indian Institute of Technology, Delhi
    Adversarial attacks on Graph Neural Networks

Awards

Interests

Reliability and alignment measurement
Adversarial Robustness
Structural Alignment
Reliable and interpretable predictive modeling
Structure-inspired self-interpretable models
Robust training schemes
User-controllable generation
Graph Generation under convex constraints
Controllable text generation

References

Prof. Srijan Kumar
Assistant Professor, Georgia Tech
Dr. Rakshit Trivedi
Postdoctoral Associate, MIT
Prof. Sayan Ranu
Associate Professor, IIT Delhi
Prof. Sourav Medya
Assistant Professor, UIC