Romina Aalishah
PhD Student · Johns Hopkins University
Laboratory of Adaptive and Maladaptive Intelligence (LAMI) · Advisor: Ilya Monosov
🧠 Machine Reasoning · ⚡ Edge Computing · 👁 Computer Vision
I study the gap between looking intelligent and being intelligent. At Johns Hopkins, my work sits at the intersection of machine reasoning, computer vision, and model compression, investigating how neural networks learn, represent, and transfer knowledge, and what is preserved or lost when large models are made small. I'm drawn to problems where the engineering question and the scientific question turn out to be the same question.
✨ Currently collaborating closely with the best research/life partner: Arman Hatami ❤
Measuring whether distilled students preserve rule-sensitive internal representations or only mimic final outputs.
Hardware-aware efficient models using Mamba, low-rank decomposition, and KD for on-device inference at the edge.
Efficient vision models for image classification, object detection, and super-resolution on edge hardware.
Depth-aware removal of forget-specific directions for class-level unlearning in computer vision models.
Experience
News
Awards & Honors
CVPR 2026 Broadening Participation Scholarship
Received a $600 travel stipend and registration waiver to attend CVPR 2026.
2026Second Best Paper Award; AAAI 2025 Spring Symposium
For "GenAI at the Edge: Comprehensive Survey on Empowering Edge Devices," presented at the Special Session on Generative AI @ Edge.
2025Best Poster Award (People's Choice); Johns Hopkins University
For "MambaLiteSR: Image Super-Resolution with Low-Rank Mamba using Knowledge Distillation."
2025DAC 2025 Young Student Fellow & Emerging Contributor Award
Recognized at the 62nd Design Automation Conference (DAC 2025). Received a $500 travel grant and full conference registration.
2025NeurIPS 2025 Student Volunteer
Selected as a student volunteer for NeurIPS 2025, San Diego; received complimentary registration.
2025Code-Cup Programming Competition
Ranked 32nd out of 1,073 in Data Science and 68th out of 2,417 in Algorithms.
2023National Universities Entrance Exam (Iran)
Ranked 225th out of 164,000 in the Mathematics and Physics track.
2019