About Me

I'm a machine learning researcher and a recent M.S. in Computer Science graduate from UMass Amherst. I work on agentic and multimodal AI, with a focus on systems that stay reliable outside of benchmarks. I'm open to full-time applied research roles.

I believe talent and curiosity show up everywhere, even where opportunity doesn't, so I'm always happy to share advice, talk through research directions, and support students and researchers from underrepresented or less-resourced backgrounds.

My Erdős Number is 4 (Sharma–Rossi–Duffield–Alon–Erdős).

Interests

• Agentic AI & LLM agents
• Multimodal intelligence
• Embodied AI
• Robust & efficient ML
• Reinforcement learning

Publications

Research Projects

Symmetry and Lie Group Theory

Symmetry is a pivotal concept in machine learning, suggesting that a model or algorithm remains robust or invariant under specific transformations, such as rotation, reflection, or scaling of input data. Extending this principle, Lie Group Theory, which deals with continuous symmetry, finds compelling applications in the field of machine learning. Lie groups, as mathematical structures, embody sets of elements that demonstrate both algebraic and geometric properties. This unique blend enables them to effectively represent and process smooth transformations in data.

Hyperparameter Tuning using Keras Tuner

Developing machine learning models is usually an iterative process. You start with an initial design then reconfigure until you get a model that can be trained efficiently in terms of time and compute resources. As you may already know, these settings that you adjust are called hyperparameters. These are the variables that govern the training process and the topology of an ML model. These remain constant over the training process and directly impact the performance of your ML program. The process of finding the optimal set of hyperparameters is called hyperparameter tuning or hypertuning, and it is an essential part of a machine learning pipeline. Without it, you might end up with a model that has unnecessary parameters and take too long to train.

[Sep 2026] Released "Joint and Cross-Modal Video-Audio Generation and Editing: A Unified Formulation and Design Taxonomy" on arXiv, with co-authors from Adobe Research, Cisco, Dolby, Stanford, UIUC, USC, Texas A&M, Vanderbilt, Virginia Tech, Arizona State, UT Dallas, and the University of Oregon.
[Jul 2026] Presented our poster on Test-Time Strategies for More Efficient and Accurate Agentic RAG at ACL 2026.
[May 2026] Graduated from UMass Amherst with an M.S. in Computer Science.
[Apr 2026] Updated my Erdős Number to 4 via co-authorship with Ryan A. Rossi (Sharma–Rossi–Duffield–Alon–Erdős).
[Apr 2026] Test-Time Strategies for More Efficient and Accurate Agentic RAG accepted at ACL work done under Adobe.
[Sep 2025] Joined Amazon as an Applied Scientist Intern, working on large-scale seller simulation using AI systems.
[May 2025] Joined IBM Research as a Research Scientist Intern, focusing on developing efficient inference accelerator.
[Feb 2025] Started as a Graduate Student Researcher at Adobe, working on agentic RAG, iterative retrieval, and RL-based reasoning systems.
[Sep 2024] Began my M.S. in Computer Science at UMass Amherst.
[Aug 2024] Started at UMass as a Graduate Student in the CS Department.
[Jan 2024] Understanding Choice Independence and Error Types in Human-AI Collaboration accepted at ACM CHI'24.
[Aug 2023] Started as a Research Affiliate at Accessible and Accelerated Robotics Lab A²R Lab.
[Apr 2023] "PreAxC: Error Distribution Prediction for Approximate Computing Quality Control" accepted to ISQED 2023.
[Jan 2023] Started as an SDE Intern in the PV Payments team at Amazon Prime Video.
[Jun 2022] University of South Carolina Coverage: Poster Presentation as Junior McNair Fellow.
[Jun 2022] Visiting Research Specialist at University of South Carolina, USA.
[Jan 2022] Served as Reviewer for Design Automation Conference(DAC).