Nikhil Muralidhar

Bharat is a Ph.D student in his third year in the ScAI lab. His research is focused on developing machine learning techniques with a focus on improved generalization under data paucity and compute paucity contexts.  Bharat's research has been published in prestigious venues like AAAI and ICLR.

Reihaneh is a second year Ph.D student in the ScAI lab. Her research is focused on developing effective machine learning techniques in few-shot domain adaptation contexts. As part of her research, Reihaneh is exploring few-shot adaptation in contexts like wireless channel estimation as well as personalized product search.

Shital is a first year Ph.D student in the ScAI lab. His research is focused on developing effective agentic pipelines in LLMs to accelerate scientific discovery. Shital is currently exploring the effectiveness of LLMs in various aspects of the computational simulation pipeline. Additionally, Shital also leads the ScAI lab effort for flu forecasting conducted annually by the CDC as part of the Flusight competition.

Nilay is a MS student in the computer science department at Stevens Institute of Technology who conducted research for his MS thesis as part of the ScAI lab at Stevens (2023 - 2025). His research focused on alleviating catastrophic training failures, in popular physics informed neural network (PINN) models. As part of his research, Nilay developed a novel `learned initialization' strategy using which, PINNs were shown to overcome training failures and even improved in extrapolation performance, compared to randomly initialized PINNs. 

Ali is a MS student in the computer science department at Stevens Institute of Technology who conducted research for his MS thesis as part of the ScAI lab at Stevens (2023 - 2025). His research focused on developing LLM-modulo techniques to improve eCommerce search ranking under data paucity. Specifically, as part of his research, Ali explored leveraging LLMs to (i) identify ambiguous user queries, (ii) generate hard-negatives for a query as well as (iii) to generate soft-positives (i.e., variants) of user queries. All these synthetically generated queries and query hardness estimates from the LLMs were then incorporated to inform training curricula as well as influence the latent space structure via. energy-based losses of traditional deep learning based search ranking models. Overall, Ali's research highlighted the effectiveness of LLMs in significantly improving the ranking prowess of  traditional deep learning pipelines for eCommerce search ranking.  Ali, is set to start his Ph.D at Virginia Tech in Fall 2025.

Ishaan conducted research as a high-school student in our lab and led the epidemiological modeling efforts related to West Nile Virus (WNV) and Influenza-like Illness (ili) forecasting for the 2023 - 2024 season as part of CDC led competitions. Ishaan's work on developing novel forecasting techniques for WNV was also recognized as one of the top 300 submissions (out of 2,162 submissions) nation-wide for the annual Regeneron Science Talent Search competition.