cv

Master of Science, Biomedical Engineering, Research Track

• Computational focus. Completed May 2026.
• {"Relevant coursework"=>"Computer Vision, Visual Learning and Recognition, Image-Based Computational Modelling and Analysis, Learning for 3D Vision, Clinical Translations of AI."}

Bachelor of Technology

• Major in Biomedical Engineering, minor in Data Science.
• {"Relevant coursework"=>"Pattern Recognition, Digital Image Processing, Digital Signal Processing, Signals and Systems, Statistical Inference and Machine Learning."}

Staff Scientist, Surgical Biomechanics Lab

• Patient-specific vascular geometry and multiphysics modeling for geometric deep learning based predictive modeling.
• Lead computer vision and deep learning researcher under Dr. Luka Pocivavsek, investigating relationships between integrated baseline Gaussian curvature, longitudinal curvature-adaptive patch-wise surface-area growth, and CFD-derived pressure and velocity fields.
• Building CT/CTA-to-3D vascular modeling pipelines spanning segmentation, surface reconstruction, mesh processing, and anatomical correspondence, with extension toward biomechanical stress analysis.

Research Collaborator, Surreality Lab

• Enhanced intraoperative speech analysis for awake craniotomy (TRACER).
• Contributing to an automated operating-room speech pipeline combining speaker attribution, word-onset localization, overlap detection, and neural source separation.
• Evaluating overlap detection and selective source separation, auditing failure modes and robustness of diarization, temporal localization, and separation components.

Graduate Student Researcher, Computational Engineering and Robotics Lab

• Real-time 3D vascular deformation estimation from monocular fluoroscopy for endovascular robotic surgery.
• Developed FluoroGRID, a multimodal deep learning framework combining fluoroscopic input with preoperative vascular geometry, sparse control points, and acquisition metadata to predict 3D deformation.
• Designed a quad-encoder architecture fusing image features, graph representations, sampled mesh geometry, and camera metadata with differentiable projection and physics-based geometric regularizers.
• Built a synthetic data pipeline generating over 370,000 paired 3D-2D training examples; achieved real-time inference with sub-millimeter reconstruction precision. Presented initial results at the ELSR workshop, IROS 2026.

Research Assistant, The ∀ Lab and Image Science Lab

• Longitudinal glioblastoma evolution and MRI-based forecasting.
• Built an Attention U-Net with temporal embeddings and Neural ODE dynamics for continuous-time glioblastoma forecasting from longitudinal multimodal MRI.
• Reduced prediction MSE by 6.7% versus a persistence baseline and relative FLAIR-volume error by 98%; presented at the 2025 CMU Biomedical Engineering Forum.

Graduate Teaching Assistant

• Supported graduate courses in Applied Deep Learning, Fundamentals of Computational Biomedical Engineering, Computer Vision for Engineers, and Machine Learning in Experimental Biomedical Engineering Research.

Sparse-View Cardiac Reconstruction with Neural Implicit Representations

• Coordinate-based neural implicit reconstruction completing dense 3D left-ventricular geometry from three sparse 2D echocardiography sections, scored only on unobserved voxels, achieving 0.936 Dice and 0.880 IoU on MITEA.
• Matched comparison against an explicit Gaussian occupancy field under identical observations (0.902 Dice, 0.821 IoU), with mesh and topology diagnostics distinguishing genuine shape error from label artifacts.

Deformable 3D Reconstruction of Live Endoscopic Surgical Scenes

• Prototyping deformable 3D Gaussian Splatting for dynamic monocular endoscopic reconstruction with depth/pose initialization, differentiable rendering, and deformation fields.
• Exploring semantic feature alignment for open-vocabulary querying of reconstructed surgical scenes.

Geometric Multi-View Breast Cancer Detection

• Geometric alignment and dual-view correspondence framework for paired CC/MLO mammography, integrating lesion segmentation, localization, registration, and cross-view attention.
• Reduced cross-view lesion-centroid discrepancy by 99.1%, enabling spatially aligned multi-view representations for malignancy classification.

Sparse-View Coronary Vessel Reconstruction

• Research prototype for sparse-view 3D coronary vessel reconstruction from angiography projections using Gaussian geometry, differentiable rendering, and PINN-based hemodynamic regularization. Now archived.

Open Horizon Robotics Perception Curriculum

• Authoring open-source computer vision curriculum covering classical vision, deep learning, 3D vision, geometry, and perception physics.

The use of artificial intelligence and machine learning to predict tumor recurrence in high-grade gliomas - a systematic review

Toward real-time 3D vascular state estimation from monocular fluoroscopy for endovascular robotic neurosurgery

A review of robotics in achieving autonomous and remote neuroendovascular intervention

Automated histopathological detection and classification of lung cancer with an image pre-processing pipeline and spatial attention with deep neural networks

An explainable artificial intelligence integrated system for automatic detection of dengue from images of blood smears using transfer learning

Deep learning based detection of monkeypox virus using skin lesion images

Detection of monkeypox from skin lesion images using deep learning networks and explainable artificial intelligence

Processing and detection of lung and colon cancer from histopathological images using deep residual networks

Real-time 3D neuro-vascular deformation estimation from sparse 2D angiography

Sparse-view cardiac reconstruction with neural implicit representations

• Biomedical Engineering Department Head's Fellowship, Carnegie Mellon University

• Best Paper, AI Track, Second International Conference on Artificial Intelligence, Computational Electronics and Communication