My research focuses on understanding the limitations and capabilities of contemporary AI systems, such as Large Language Models (LLMs), through mechanistic interpretability and model behavior control techniques like model editing and representation engineering. I am particularly interested in the mechanisms behind LLM behaviors and the democratization of these models by ensuring they are secure and unbiased without extensive fine-tuning. My work at Microsoft involved training mid-sized language models like Phi and Turing Bletchely in multilingual and multimodal settings respectively and integrating them into products like Bing Search and Microsoft Copilot.
Keywords: LLMs, model interpretability, relational compositionality, activation steering, SafeAI, multimodal learning, bias mitigation, multilingual learning
I graduated with a Bachelor's degree in Electrical Engineering, and minor in Computer Science, from IIT Kharagpur in 2022. For more detailed information about my background, please refer to my CV. If you’d like to chat about my work or research interests, feel free to reach out!
Joined Multisensory Intelligence Research Group at MIT Media Lab!
Joined Harvard University and DtaK Lab!
Our paper on the effect of Multilingualism on Value Pluralism in LLMs has been accepted in LREC-COLING 2024!
Our paper on Foreign Language Effect in LLMs through moral philosophical lens has been accepted as Oral in EACL 2024!
Our paper on Value Pluralism in LLMs has been accepted in Findings of EMNLP 2023!
Our paper on Visual Document Understanding has been accepted in EMNLP Industry Track 2023!
Taught a workshop course in Practical NLP and Large Language Models (LLMs) for bachelor's students at the Indian Institute of Science (IISc) as part of Kotak-IISc AI-ML Centre!
Joined Microsoft Research India as a Research Fellow.
sPhinX: Sample Efficient Multilingual Instruction Fine-Tuning Through N-shot Guided Prompting
Sanchit Ahuja*, , Hardik Hansrajbhai Chauhan, Barun Patra, Kriti Aggarwal, Luciano Del Corro, Arindam Mitra, Tejas Indulal Dhamecha, Ahmed Awadallah, Monojit Choudhary, Vishrav Chaudhary, Sunayana Sitaram
Under Review pdf
Do Moral Judgment and Reasoning Capability of LLMs Change with Language? A Study using the Multilingual Defining Issues Test
Aditi Khandelwal*, Utkarsh Agarwal*, , Monojit Choudhury
EACL 2024 Oral pdf
Ethical Reasoning and Moral Value Alignment of LLMs Depend on the Language we Prompt them in
Utkarsh Agarwal*, , Aditi Khandelwal*, Monojit Choudhury
LREC-COLING 2024 pdf
Probing the Moral Development of Large Language Models through Defining Issues Test
, Aditi Khandelwal*, Utkarsh Agarwal*, Monojit Choudhury
Workshop on AI meets Moral Philosophy and Moral Psychology (MP2) - Neurips 2023 pdf
Ethical Reasoning over Moral Alignment: A Case and Framework for In-Context Ethical Policies in LLMs
Abhinav Rao*, Aditi Khandelwal*, , Utkarsh Agarwal*, Monojit Choudhury
Findings of EMNLP 2023 pdfslidesvideoposter
DUBLIN: Visual Document Understanding By Language-Image Network
Kriti Aggarwal*, Aditi Khandelwal*, , Owais Mohammed Khan, Qiang Liu, Monojit Choudhury, Hardik Hansrajbhai Chauhan, Subhojit Som, Vishrav Chaudhary, Saurabh Tiwary
EMNLP Industry Track 2023 pdfslidesvideoposter
Augmented Reality Based Recommendations based on Perceptual Shape Style Compatibility with Objects in the Viewpoint and Color Compatibility with the Background
, Kumar Ayush*.
Advances in Image Manipulation Workshop (AIM) - ICCV 2019 pdf
Efficient Poverty Mapping from High Resolution Remote Sensing Images
Kumar Ayush*, Burak Uzkent*, , Marshall Burke, David Lobell, Stefano Ermon
AAAI 2021 Oral pdf
Efficient Conditional Pre-training for Transfer Learning
Shuvam Chakraborty, Kumar Ayush*, Burak Uzkent*, , Evan Sheehan, Stefano Ermon
Workshop on Learning with Limited Labelled Data for Image and Video Understanding (L3D-IVU) - CVPR 2022 pdf
Geography-Aware Self-Supervised Learning
Kumar Ayush*, Burak Uzkent*, Chenlin Meng*, , Marshall Burke, David Lobell, Stefano Ermon
ICCV 2021 pdf
