AI/ML research scientist building reliable intelligent systems for noisy, multimodal physical-world data.
I work at the intersection of scientific machine learning, computer vision, probabilistic inference, Physical AI, and agentic systems.
-
GeoWorld Studio — typed, reproducible geoscience workflows that turn structured scenarios into synthetic earth models, physical properties, reflectivity, and seismic responses.
Engineering blog -
SnapGraph3D — constraint-aware 3D assembly with scene graphs, validation, repair, and deterministic benchmarking.
-
JobTracker — application tracking and transparent job-skill analysis built with React, Express, Playwright, and local LLM workflows.
- SAGE-AVO: structure-aware multimodal inversion with transformers, dynamic graph learning, conditional flow matching, and physics-based loss functions.
- TLNet: 3D/4D change detection and segmentation under sparse labels.
- Probabilistic inference and uncertainty quantification for nonlinear inverse problems.
Python · PyTorch · FastAPI · Pydantic · PostgreSQL · AWS · GPU/HPC
Scientific ML · Computer Vision · Generative Models · Physical AI · Agentic Systems