Research
Interested in embodied real-time intelligence. My collaborators and professors are great! :)
Publications
- Learning to Assist: Collaborative VLAs for Implicit Human-Robot Collaboration — Submitted to CoRL 2026, 1st Author (Jun 2026)
- SKiM-GPT: Biomedical Literature-Based Discovery with LLM Hypothesis Evaluation — BMC Bioinformatics, 3rd Author (Jul 2025)
- Susceptibility of Adversarial Attacks on Medical Image Segmentation Models — 2023 ISBI, Shared 1st Author (Feb 2023)
- Deeply Supervised Transformer Encoders — Independent Research in High School (Feb 2023)
Research Experience
Robot Teaching & Teaming Lab | Undergraduate Researcher
- Research under Prof. Mike Hagenow
- Trained real-time collaborative policies for assembly (handover, turn-taking); designed fixtures in Onshape
- Sped up VLA inference from 140 ms to 70 ms via KV-cache broadcasting and AdaRMSNorm tabulation in JAX
- Built lab-wide ROS 2 robot stack: joint-velocity P-controller, IK, spacemouse teleop, used by 11 researchers
- Engineered 80 Hz multimodal pipeline for streaming RGB, point clouds, and robot state with real-time visuals
- Increased closed-loop robot motion speed by 1.6× while maintaining low trajectory-tracking error
- Reduced premature collaborative actions by 6× via inference-time steering algorithm
- Built unified training, evaluation, and real-time inference infrastructure for four diffusion and VLA policy families
ADAPT @ UIUC | Undergraduate Researcher
- Research under Prof. Charith Mendis
- Desugared 46 TASO rewrite rules into XLA-HLO in TensorRight DSL (verifies ML compiler tensor rewrites)
- Expanded DSL with rank lower-bound constraints to enable both rank monomorphic and polymorphic verification
Stewart Lab @ Morgridge Institute for Research | Research Intern
- Supervised under Ron Stewart @ Morgridge Institute for Research
- Published paper in BMC Bioinformatics
- Combined biomedical literature-based discovery with LLM hypothesis evaluation (feel free to play with it here)
- Fine-tuned LLM for biomedical text analysis with synthetic data (9,000+ downloads on HuggingFace!)
- Improved accuracy of RAG relevance filter from 70% to 90% with NEFTune and rsLoRA
- Sped up inference pipeline 15× via packed batching + vLLM on university HTC cluster
Susceptibility of Adversarial Attacks on Medical Image Segmentation Models | Independent Publication in High School
- Self published at IEEE's 2023 International Symposium on Biomedical Imaging (ISBI) conference. I got cited twice! 🎉
- Explored the efficacy of adversarial attacks on SOTA image segmentation models (UNet, UNet++, UNet + ResNeXt-101 backbone, UNet + EffNet-B7 backbone)
- Discovered that varying the loss used in FGSM improved Attack Success (defined in paper)
Deeply Supervised Transformer Encoders | Independent Research in High School
- Supervised under Prof. Mohammad Taher Pilehvar @ University of Cambridge
- Examined the application of UNet++ style deep supervision in transformer encoders, which unfortunately didn't work very well :(
- Adding Deep Supervision increased BERT accuracy by 5% on the WNLI benchmark