I work at the intersection of neuroscience, data science, and software engineering. My focus is turning complex experimental data into analysis pipelines that are understandable, testable, and useful to researchers.

I care about:

• quantitative models that make assumptions explicit;
• reliable workflows for high-dimensional and time-series data;
• validation that respects the experimental design; and
• documentation that makes research software easier to trust and reuse.

Languages & analysis
Python · SQL · NumPy · pandas · SciPy · scikit-learn

Scientific computing
Neural time series · electrophysiology · behavioral data · statistical modeling · reproducible research

Engineering
Data validation · nested cross-validation · testing · command-line tools · technical documentation

• neural population dynamics and electrophysiology;
• interpretable machine learning for biomedical datasets;
• adaptive and closed-loop modeling of biological systems; and
• better interfaces between scientific analysis and usable software.

• Latent-period epileptogenesis via Markov modeling — hidden-Markov modeling of larval-zebrafish LFP after traumatic brain injury.
• Behavioral PTE — visual-habituation kinetics as a candidate predictor of later seizure-like activity.
• Decision geometry (IBL Brain-Wide Map) — population decoding and cross-temporal dynamics in real IBL Neuropixels recordings.
• CoRe-TBI — within-animal counterfactual recovery modeling of movement after traumatic brain injury.
• Lean Proof Diagnostics Explorer — transparent failure analysis of Lean proof-attempt logs.
• Autoformalization and iterative proof repair — a Lean generation/verification loop with language-model proof repair.
• Gravitational-wave inference engine — a dependency-light gravitational-wave inference teaching pipeline.
• Astro transient pipeline — simulated multi-messenger transient triage, from alert stream to follow-up queue.
• Planetary autonomy sim — seeded terrain generation and cost-based rover route planning.
• Multilayer perceptron from scratch — a NumPy-only MLP with backpropagation, calibration metrics, and tests.
• PharmaBlast — physics-informed modeling of fluorescence-based detection of pharmaceutical-laden microplastics in water.

Question → measurable endpoint → leakage-safe analysis → reproducible result

I prefer small, well-tested analytical steps over opaque complexity. When a model is not supported by the data, I document that limitation clearly and use it to guide the next experiment.

If you are working on computational neuroscience, biomedical data, or research software, I would be glad to connect.

LinkedIn · Repositories · Email