A research team working on connectomics and brain simulation needed publication-quality figures from complex, multi-dimensional datasets. I was the visualization specialist across four phases, from initial prototypes to the final publication-ready figures, working with messy and inconsistent data throughout.
Highlights
- Created multi-scale visualizations of neuron counts in simulations with biological reference lines for real organisms (mouse, human), spanning many orders of magnitude with appropriate log-scale transformations.
- Developed complex heatmaps revealing patterns in neural recording capabilities and simulation coverage across methods and brain regions.
- Built innovative 3D visualizations mapping organism computational requirements across multiple dimensions using matplotlib's 3D toolkit.
- Designed plots of storage cost evolution and brain research funding with predictive trendlines and strategic highlighting of key patterns for scientific communication.
- Wrote custom data-cleaning pipelines that extracted quantitative information from complex text descriptions using parsing techniques and AI-assisted categorization.
- Maintained organized Jupyter notebooks with well-documented data processing steps for scientific reproducibility across all four project phases.
Technology
- Python
- pandas (data transformation)
- matplotlib (custom visualization, 3D toolkit)
- seaborn
- statsmodels (regression analysis)
- NumPy
- Jupyter notebooks
- GitHub