Anjishnu Mukherjee

Pronunciation: /ʌnˈdʒɪʃnuː/ (un-JISH-noo)

I am currently on the 2027 job market for postdoctoral and tenure-track positions.

amukher6 at gmu dot edu

I am a fifth-year Ph.D. candidate in Computer Science at George Mason University, advised by Antonis Anastasopoulos and Ziwei Zhu in the GMU NLP Lab.

Research directions

My research focuses on multilingual and multicultural large language models (LLMs), and on how to find and fix their cultural failures: (1) evaluating where LLMs, vision-language models, and AI agents fail across languages and cultures, (2) studying how pretraining and instruction-tuning training data decide what a model knows about each culture, and (3) mitigating social bias in LLMs, during instruction tuning and at inference.

  1. Evaluating multilingual and multicultural LLMs

    Where do LLMs, vision-language models, and AI agents fail across languages and cultures?

    1. Global Voices (EMNLP ’23)
    2. Tower of Babel (COLM ’26)
    3. Crossroads (WACV ’25)
  2. Pretraining and instruction-tuning data

    How do pretraining and instruction-tuning data decide what a model knows about each culture?

    1. Double Trouble (EMNLP ’26)
    2. MAPLE (EMNLP ’26)
    3. Global Gallery (NAACL ’24)
  3. Mitigating social bias

    How can we mitigate social bias in LLMs, during instruction tuning and at inference?

    1. Breaking Bias (AIES ’24)
    2. KnowBias (ICML ’26)
    3. Purdah and Patriarchy (TrustNLP ’26)

Share copies a publication card image when your browser allows clipboard images; otherwise it copies the publication link.

Updates

    Selected Publications

    Selected peer-reviewed research. Full list on Google Scholar

    Evaluating multilingual and multicultural LLMs

    Where do LLMs, vision-language models, and AI agents fail across languages and cultures?

    Pretraining and instruction-tuning data

    How do pretraining and instruction-tuning data decide what a model knows about each culture?

    Mitigating social bias

    How can we mitigate social bias in LLMs, during instruction tuning and at inference?

    Mentees

    High school

    1. Lauren Chen

      Syosset High School

      Multilingual and multi-attribute cultural steering for LLMs

    2. Hasini Jasthi

      Thomas Jefferson (TJHSST)

      Accent and dialect disparities in voice assistants, focusing on South and Southeast Asian English

    Undergraduate

    1. Anubhav Maity

      IIT Roorkee · Qualcomm

      Som Shekhar Sharma

      IIT Roorkee · InMobi

      Akshat Sharma

      IIT Roorkee · Glance

      Cultural adaptation for Indian food images (B.Tech thesis) · project

    Graduate · M.S.

    1. Mamnuya Rinki

      GMU · IntraFi

      South Asian bias in multilingual, open-ended LLM generations (M.S. thesis) · paper

    2. Abhi Krishna

      GMU · SyncoreLabs

      Auditing cultural representation in text-to-image models (AI, Power & Society course project) · report

    3. Aksh Patel

      GMU · Aireon

      Intersectional bias across Indo-Aryan languages (Advanced NLP course project) · slides

    4. Diwita Banerjee

      GMU · DoorDash

      Assessing cuisine biases and detecting hallucinations in chart understanding in LLMs

    Teaching

    CS 112 · Introduction to Python Programming

    Teaching Assistant · George Mason University

    Talks & Service

    Talks and presentations

    1. MAPLE · Lightning talk, GMU School of Computing PhD Research Symposium
    2. LLM agents · Guest lecture, CS 478 (Natural Language Processing), George Mason · slides
    3. MAPLE: Metadata Conditioned Pre-training for Locale Aware Question Answering · EMNLP oral presentation, Budapest · slides
    4. Measuring and Modeling Cultural Representations · Guest lecture, CS 678 (Advanced NLP), George Mason · slides
    5. Crossroads of Continents · WACV oral presentation, Tucson · slides
    6. Diffusion Models · Guest lecture, CS 678 (Advanced NLP), George Mason · slides
    7. Multilingual socio-cultural biases · University of Toronto invited speaker series · Remote · slides
    8. Global Gallery · MASC-SLL at Johns Hopkins · slides
    9. Cross-Lingual Biases and Cultural Understanding in LLMs · Notre Dame NLP seminar · Remote · slides

    Academic service

    1. Reviewer · COLM · AIES
    2. Reviewer · ACL Rolling ReviewACL · EMNLP · NAACL · EACL
    3. Sub-reviewer · COLM · AIES
    4. Workshop reviewer · AfricaNLP · DravidianLangTech · EACL SRW · NewInML · WiNLP · LT-EDI
    5. Journal reviewer · ACM TIST

    Awards

    1. Outstanding Paper AwardEMNLP 2026, for Double Trouble
    2. Distinguished Academic Achievement AwardGeorge Mason University
    3. Outstanding GTA AwardGeorge Mason University
    4. Outstanding ReviewerEMNLP
    5. Summer Graduate Research Award (Awarded $8,500)George Mason University
    6. DAAD WISE Scholarship (Awarded $3,000)German Academic Exchange Service