Hello! I’m Arka, a second-year Computing and Information Sciences PhD student at the Rochester Institute of Technology’s Thomas B. Golisano College of Computing and Information Sciences. I am a part of the Social Insight Lab, where I am advised by Professor Ashiqur R. KhudaBukhsh. My research focuses on the intersection of Natural Language Processing and Computational Social Science. Specifically, I combine classical Machine Learning and modern NLP theories, and Explainability techniques to develop fair, responsible, and equitable Generative AI systems. My work has been recognized at top AI conferences, including AAAI and IJCAI, and has attracted media attention, such as a feature article in the Montreal AI Ethics Magazine and the WIRED. I am also passionate about mathematics, regularly engaging with advanced topics such as combinatorics, optimization, probability, and number theory. I strive to integrate this mathematical passion into my research and further explore my growing interest in Quantitative Finance.
Latest Updates
- Our work on bias auditing LLMs featured in the WIRED in relation to socio-political consequences. [Article]
- Our work on Jailbreaking and bias-auditing got accepted as Student Abstract in AAAI 2025 . [Article]
- RIT News featured our work on bias auditing LLMs. [Article]
- Montreal AI Ethics featured our work on bias auditing LLMs. [Article]
- Awarded the RIT Language Science Student Excellence Award 2024 for research in NLP
Selected Publications
- Down the Toxicity Rabbit Hole: A Novel Framework to Bias Audit Large Language Models. [PDF] (IJCAI 2024).
Skills
- Machine Learning (ML), Artificial Intelligence (AI), Natural Language Processing (NLP), Large Language Model (LLM), Explainable AI (XAI), Reinforcement Learning with Human Feedback (RLHF), Deep Learning, Responsible AI, Quantitative Finance, Optimization, Statistical Inference, Quantitative Research, Predictive Modeling, Data Visualization, Leadership.
- Python, C/C++, MATLAB, SQL, Transformers, PyTorch, TensorFlow, NumPy, Pandas, NLTK, Spacy, HuggingFace, SHAP
Selected Achievements
- RIT Language Science Student Excellence Award (2024): One of the only two recipients of the award among 1000+ students university-wide in 2024; for the research in responsible AI and NLP systems.
- Common Admission Test (CAT, 2022): Ranked 98.97 percentile in highly competitive CAT 2022 exam held among 100k+ students in India as premier B-School admission test. Offered an admission from Indian Institute of Management, Indore (IIMI) PGP programme; one of the most premier B-Schools in the world.
- Regional Mathematics Olympiad (2018): Qualified for prestigious RMO, WB; demonstrating advanced level aptitude in Discrete Mathematics, Combinatorics, Number Theory, Algebra, Geometry, and Optimization.
- Jagadish Bose National Science Talent Scholarship (2017): Awarded prestigious JBNSTS scholarship for excellence in science