About
I am a computational physicist finishing my PhD at UC Santa Cruz, where my research focuses on black hole dynamics and gravitational waves. My work sits at the boundary between theory and implementation: 5 peer reviewed publications (4 first author), cross institutional collaborations at ADMX and Fermilab, and end to end pipelines spanning simulation, signal processing, and ML systems. My technical work centers on building frameworks that expose where models and predictions break down. That includes matched filtering systems for weak signal detection, evaluation pipelines that characterize failure modes across thousands of noise realizations, and simulation tools that extend existing codebases into new physical regimes. Outside of physics, I won top project at the Erdős Institute among roughly 200 PhD candidates, building a 3D deep learning pipeline for MRI brain tumor segmentation from scratch — Dice = 0.83, 86% precision on held-out data.
I am actively looking for research scientist and research engineer roles in the SF Bay Area.
Research Software and Analysis Pipelines
ADMX Gravitational Wave Detection Pipeline
End to end signal analysis system built in collaboration with ADMX and Fermilab to search for gravitational waves using resonant microwave cavity detectors.
- Matched filtering pipelines across multi-terabyte experimental datasets, turning a manual analysis workflow into a reproducible, automated one
- Frequency domain signal processing including FFT binning, PSD estimation via Welch and robust stacking, and receiver response calibration
- Synthetic signal injection pipelines that stress tested detection models across thousands of noise realizations, surfacing failure modes and producing concrete evaluation criteria
- Open source code on GitHub
Black Hole Simulation Framework
Extension of a widely used open source black hole simulation codebase to a physical regime it was never designed to handle.
- Identified the mathematical assumptions bounding the original codebase and overhauled its core dynamics to model charged charge depletion, producing the first computational framework of its kind for this class of problem
- Covers both spin evolution and an approximate method for the time evolution of charged black holes, extending the original code which only handled spinning or static configurations
- Numerical evolution pipelines covering ODE solvers, parameter sweeps, and interpolation across physically motivated initial conditions
- Results published as a first author paper in Physical Review D
- Open source code on GitHub
3D Brain Tumor Segmentation
End to end deep learning pipeline for volumetric MRI segmentation, awarded top project among roughly 200 PhD candidates at the Erdős Institute.
- Achieved Dice = 0.83 and 86% precision on held out test data across 500 patients and 4 MRI modalities
- Designed a composite loss function combining BCE and Dice to address severe class imbalance between tumor and non-tumor regions
- Built the full pipeline from data ingestion through training, threshold optimization, and inference evaluation using GPU accelerated PyTorch U-Net architectures
- Open source code on GitHub
Recent publications
-
Quantum Tunneling of Primordial Black Holes to White Holes: Rates, Constraints, and Implications for Fast Radio Bursts
-
Dark-sector modifications to Kerr and Reissner–Nordström black hole evaporation
-
Constraints on the maximal number of dark degrees of freedom from black hole evaporation, cosmic rays, colliders, and supernovae
-
Precision gravity constraints on large dark sectors
-
Maximal gravitational wave signal from asteroid-mass primordial black hole mergers at resonant microwave cavities
Technical Expertise
Signal Processing and Detection
- Matched filtering and optimal detection statistics for weak signal recovery
- Power spectral density estimation (Welch, median stacking, robust averaging)
- Frequency domain analysis, FFT binning, receiver response calibration
- Noise modeling, stationarity testing, and false positive characterization
Scientific Computing and Simulation
- Large scale simulation pipelines spanning parameter sweeps and ODE evolution
- Monte Carlo methods, uncertainty propagation, and sensitivity estimation
- Extending existing research codebases to new physical regimes
- Reproducible benchmark design and systematic evaluation frameworks
Machine Learning and Evaluation
- End to end deep learning pipelines in PyTorch (CNNs, 3D U-Net architectures)
- Loss function design, class imbalance handling, and domain dataset curation
- Rigorous model evaluation across Dice, IoU, ROC, and threshold optimization
- Synthetic data pipelines that stress test models across distribution shift
Programming and Tools
- Python: NumPy, SciPy, pandas, PyTorch, PyCBC, Matplotlib
- C and C++, FFTW, HDF5, Git, Linux
- Multi-terabyte data pipelines and HDF5 based workflows
- LaTeX, technical documentation, and open source research code
Contact
Best way to reach me is email. Currently finishing my PhD at UC Santa Cruz (expected 2026) and open to research scientist and research engineer roles in the SF Bay Area.