Zahra Safdari Fesaghandis

Zahra Safdari Fesaghandis

M.Sc. Student · Research & Teaching Assistant · ASAP Lab, Department of Computer Engineering

Bilkent University · Ankara, Türkiye · Expected graduation: July 2027

I work on trustworthy AI from two complementary directions: privacy and security of sensitive data, and socially responsible language technologies. My current research includes genomic privacy and membership-inference defenses, multilingual hate speech and counterspeech, and multimodal counterspeech generation.

Privacy-Preserving ML Genomic Privacy Social Computing Multilingual NLP Multimodal Counterspeech Trustworthy AI
Publications Download CV Contact
01

About

I am a Master's student in Computer Engineering at Bilkent University (Sep 2025–Expected Jul 2027), where I work as a Research & Teaching Assistant in the ASAP Lab under the supervision of Dr. Sinem Sav.

At Bilkent, my research focuses on genomic privacy, membership-inference attacks, and defenses against colluding adversaries in genomic beacon systems. I am also a core graduate researcher on a TÜBİTAK 1001-funded project on dynamic privacy defenses for genomic data sharing.

In parallel, I continue to work with Prof. Suman Kalyan Maity at Missouri University of Science and Technology on social computing and multilingual NLP. My current work extends my previous text-based counterspeech research toward image-based and multimodal counterspeech generation for combating online hate.

I received my B.Sc. in Computer Science from the University of Tabriz (Sep 2020–Sep 2024), where I ranked 3rd in my class. I also completed a research internship at the Institute for Research in Fundamental Sciences (IPM).

Research Interests
  • Privacy-Preserving Machine Learning
  • Genomic Privacy & Membership Inference
  • Social Computing & Computational Social Science
  • Hate Speech & Counterspeech
  • Multilingual & Low-Resource NLP
  • Multimodal NLP & Vision-Language Models
  • Trustworthy & Responsible AI
02

Publications & Manuscripts

01
Published
ParsCN: A Persian Dataset and Framework for Counter-Narrative Generation to Combat Hate Speech
Zahra Safdari Fesaghandis, Suman Kalyan Maity
ICWSM 2026
A culturally grounded Persian counter-narrative resource with 1,100 hate speech–counter-narrative pairs spanning six target groups and six countering strategies, built through human annotation, multilingual adaptation, LLM-assisted generation, and human curation.
Project & Dataset Google Scholar
02
Minor Revision
Multilingual Hate Speech Detection and Counterspeech Generation: A Comprehensive Survey and Practical Guide
Zahra Safdari Fesaghandis, Suman Kalyan Maity
ACM Computing Surveys — under minor revision
A comprehensive survey of multilingual hate-speech detection and counterspeech generation, with emphasis on multilingual transfer, low-resource settings, model evaluation, practical design choices, and responsible deployment.
03

Research & Projects

🧬
Genomic Privacy & Beacon Defense
Investigating membership-inference attacks, coordinated multi-account adversaries, and adaptive privacy defenses for genomic beacon systems while studying the privacy–utility trade-off in sensitive data sharing.
🖼️
Image-Based Counterspeech Generation
Ongoing research with Prof. Suman Kalyan Maity on multimodal methods for generating context-aware counterspeech in response to hateful image-based content on social media.
🌐
ParsCN — Persian Counter-Narrative Dataset
Developed a Persian counter-narrative dataset of 1,100 hate speech–counter-narrative pairs across six target groups and six strategies using human annotation, culturally adapted multilingual resources, semantic retrieval, LLM-assisted generation, and human curation.
📚
FRAG — Fact-Grounded Counterspeech
Investigated retrieval-augmented generation for producing evidence-grounded and explainable counterspeech, connecting factual retrieval with safer social-language interventions.
💬
Multilingual Hate Speech Detection
Explored multilingual hate-speech detection across multiple languages using neural and transformer-based models, with emphasis on cross-lingual and low-resource settings.
🌍
SentiGerman
Led a team at IPM developing a German sentiment-analysis system, including data collection, model development, evaluation, and presentation.
⚖️
Fair & Explainable ML
Developed predictive models emphasizing fairness, bias mitigation, and explainability, strengthening my broader interest in trustworthy and responsible AI.
View all projects on GitHub
04

Curriculum Vitae

My full CV includes details on education, research experience,
publications, technical skills, and certifications.

↓  Download Full CV (PDF)
05

Teaching

Course Title & Role Term
CS 125
Bilkent University
 TAIntroduction to Data Analysis for Social Sciences Summer 2026
CS 115
Bilkent University
 TAIntroduction to Programming in Python Summer 2026
CS 223
Bilkent University
 TADigital Design Spring 2026
CS 223
Bilkent University
 TADigital Design Fall 2026
AI Workshop
IPM Institute
 TAArtificial Intelligence Summer Workshop Summer 2023
CS Basics
Univ. of Tabriz
 Head TABasics of Computer and Programming 2021 – 2024
Data Struct.
Univ. of Tabriz
 TAData Structures & Advanced Programming · also TA for Specialized Language; lead tutor of Python Programming Workshop 2021 – 2024
CS Courses
Meshkat Holdings
 TutorData Structures & Algorithm Design 2021 – 2022
06

Contact

I welcome inquiries about research collaborations, PhD opportunities, and questions about my work in privacy-preserving machine learning, social computing, and responsible NLP. Please feel free to reach out by email or through any of the platforms below.

✉
Email
zahra.safdari@bilkent.edu.tr
Google Scholar
Zahra Safdari Fesaghandis
LinkedIn
zahra-safdari-s81
GitHub
zahrasafdari
arXiv
arXiv
Zahra Safdari Fesaghandis

Send a Message