I build evaluation systems that expose real failure modes in AI and ship reproducible, decision-grade analytics.
Background: PhD in Theoretical Biophysics (stochastic systems, MLE, information theory), defense T&E (radar), and health-data pipelines.
Focus: AI evaluation • stochastic/dynamical systems • health/defense analytics
Contact: inwogba1@alumni.jh.edu · LinkedIn · Google Scholar
- Designed training data for LLMs, surfacing ≥75% failure on multi-step reasoning; created gold standard rubrics.
- Built end-to-end health-data pipelines, from intake and labeling to bias checks and model evaluation, with ablation studies and auto-generated, reproducible reports.
- Published paper-backed simulations of cellular electrotaxis (Soft Matter, Phys. Rev. E, Biophysical Journal); open, runnable code.
-
Predicting Short Term Traffic Delays (in progress)
Data science group project using neural networks.
Repo → https://github.com/Erdos-Projects/fall-2025-predicting-short-term-traffic-delays -
Cell-shape-and-orientation-control-galvanotactic-accuracy Soft Matter Journal companion
One-command repro of a key figure; shows shape-dependent sensing logic.
Repo → https://github.com/Nnanna100/Cell-shape-and-orientation-control-galvanotactic-accuracy -
Physical-limits-on-galvanotaxis Phys. Rev. E companion
MLE + Fokker–Planck inference demo.
Repo → https://github.com/Nnanna100/Physical-limits-on-galvanotaxis -
Keratocyte-dynamics Biophysical Journal companion Oscillation vs circling boundary; reproducible plot + short mp4/GIF.
Repo → https://github.com/Nnanna100/Coupling-cell-shape-and-velocity-leads-to-oscillation-and-circling-in-keratocyte-galvanotaxis
- BioRXiv (2025): https://doi.org/10.1101/2025.04.09.647627
- Soft Matter (2024): https://doi.org/10.1039/d4sm00952e
- Phys. Rev. E (2023): https://doi.org/10.1103/PhysRevE.108.064411
- Biophysical Journal (2023): https://doi.org/10.1016/j.bpj.2022.11.021