Shravya Kanchi

PhD Candidate

Computer Science Department, Virginia Tech

shravya@vt.edu

Google Scholar

GitHub

LinkedIn

Resume

Hey!

I am Shravya, PhD candidate at Virginia Tech, Blacksburg. I am being advised by Dr. Daphne Yao. My research interests lie in using generative AI for the good of cybersecurity.

I have experience in using generative models (like LLMs, GANs, Diffusion, etc.) for tabular data generation specific to security tasks. I also have used generative models for the creation of synthetic conversational data infused with desirable traits like empathy, politeness, etc. to mitigate toxicity in chatbot customization pipelines.

As part of my master's thesis, I designed a multi-purpose access control system to manage smart home devices. I was advised by Dr. Kamal Karlapalem. In my Bachelor's honors project, I designed an ML-based malicious short URL detector.

Updates

Apr 2025: I am now a PhD candidate!
May 2024: Pleasure to have attended IEEE S&P 2024 as a coauthor of our paper on analysis of deepfake detection schemes.
Apr 2024: Won and received the IEEE S&P Student Travel Grant 2024.
Mar 2024: Presented a poster on First Look at Toxicity Injection Attacks on Open-domain Chatbots. at DMV Security Workshop 2024.
Dec 2023: Pleasure to have attended ACSAC 2023 as a coauthor of our paper on Data poisoning attacks in Dialogue based Learning (DBL) systems.
Oct 2023: Presented on Using GenAI to strengthen security defenses at VT Fall 2023 Skillshop Series - Leveraging Creative Technologies.
Oct 2023: Our research got featured in VPM News Focal Point - “Artificial intelligence: What are the risks and benefits?”

Publications

Taming Data Challenges in ML-based Security Tasks: Lessons from Integrating Generative AI
Shravya Kanchi, Neal Mangaokar, Aravind Cheruvu, Sifat Muhammad Abdullah, Shirin Nilizadeh, Atul Prakash, Bimal Viswanath
arXiv preprint, July 2025.
PDF

TuneShield: Mitigating Toxicity in Conversational AI while Fine-tuning on Untrusted Data
Aravind Cheruvu, Shravya Kanchi, Sifat Muhammad Abdullah, Nicholas Kong, Daphne Yao, Murtuza Jadliwala, Bimal Viswanath
arXiv preprint, July 2025.
PDF

An Analysis of Recent Advances in Deepfake Image Detection in an Evolving Threat Landscape
Sifat Muhammad Abdullah, Aravind Cheruvu, Shravya Kanchi, Taejoong Chung, Peng Gao, Murtuza Jadliwala and Bimal Viswanath
IEEE S&P (Oakland) 2024, San Francisco, CA, May 2024.
PDF Code and dataset Video

A First Look at Toxicity Injection Attacks on Open-domain Chatbots
Aravind Cheruvu, Connor Weeks, Sifat Muhammad Abdullah, Shravya Kanchi, Daphne Yao, and Bimal Viswanath
ACSAC 2023, Austin, Texas, December 2023.
PDF Code and dataset Video

SEBI Regulation Biography
Sathvik Sanjeev Buggana, Deepti Saravanan, Shravya Kanchi, Ujwal Narayan, Shivam Mangale, Lini T. Thomas, Kamalakar Karlapalem, Natraj Raman
WWW Workshop, Lyon, France, April 2022
PDF

A Multi Perspective Access Control in a Smart Home
Shravya Kanchi, Kamalakar Karlapalem
CODASPY, Online, April 2021
PDF Code

Education

  • Virginia Polytechnic Institute And State University (Virginia Tech)

    Ph.D. in Computer Science

    Aug 2021 - Present

    CGPA: 3.76/4

  • International Institute of Information Technology (IIIT), Hyderabad

    Masters by Thesis in Computer Science

    Jul 2018 - May 2021

    CGPA: 9/10

  • Indian Institute of Information Technology (IIIT), Sricity

    Bachelor of Technology (Honors) in Computer Science

    Jul 2014 - May 2018

    CGPA: 8.79/10

Certifications

Qualified for and completed CopyrightX, a course on US Copyright Law, conducted each year jointly by the Harvard Law School & Berkman Klein Center for Internet and Society.

Experience

Graduate Research Assistant

IIIT Hyderabad & JP Morgan and Chase Research

Jan 2021 - Jun 2021

Developed the first named-entity labeled corpus tailored to SEBI regulations, encompassing 7,500 sub-regulations. Introduced 7 unique entity types specific to the Indian securities regulatory framework and created an Overlapping Named Entity Recognition tool with a precision of 87.47%.

Skills

GenAI expertise: LLMs (LLAMA2, FALCON, Vicuna, FLAN, OPT), Model customization (standard fine-tuning, LoRA fine-tuning), Safety alignment (Supervised fine-tuning, Direct preference optimization), Adversarial attacks, Hyper-parameter search, Prompt Engineering, AutoML

Security expertise: ML-based malware detection, Phishing, Spam, Network IDS, concept drift, BGP, website privacy, access control system, malicious URL detection

Machine Learning libraries: Huggingface, Transformers,Tokenizers, PyTorch, Numpy, Scikit-Learn, Pandas, SpaCY, RayTune, LightGBM, AutoGluon

Programming Languages: Python, C, C++, HTML/CSS

Developer Tools: SQL Developer, VS Code, Weka, label-studio, SPSS, Nmap, Docker, Jupyter Notebook, Git, Linux, Vim