Harrison Christophersen

I'm finishing an M.S. in Data Science at Gonzaga University (May 2027) and looking for data science and machine learning roles.

My research is in explainable AI. I'm first author on a paper under review at IEEE ICNC 2027 that evaluates six explanation methods across five network intrusion detectors, from random forests to graph neural networks, and shows that methods can disagree sharply while still being faithful to the model. I'm now building hybrid speech emotion recognition models for an affective computing project.

Alongside school, I launched GreaterGood.com's Amazon marketplace channel and grew it from zero to $1M in annual revenue, running split tests and building the reporting behind day-to-day decisions. Before that, I built a Python CLI at Invent.us that migrated Apache Superset dashboards between client environments in seconds.

I like problems where the analysis ends in a decision, and I care whether a result is actually true before anyone acts on it. Based in Bellevue, WA.

Research Assistant, Explainable AI at Gonzaga University (2026-05 – Present)

• Benchmarked six explanation methods, including SHAP and LIME, across five intrusion detectors from random forest to graph neural network on a 20M-flow dataset deduplicated against train/test leakage
• Showed the methods disagree substantially (top-5 feature agreement 0.19 to 0.67 vs. 0.11 chance) yet all beat random baselines on faithfulness, across 180 Bonferroni-corrected significance tests
• Found two valid criteria rank LIME best and worst, traced the reversal to correlated features, and showed a linear audit misses three quarters of the data's 72 redundant feature pairs

Research Assistant, Affective Computing at Gonzaga University (2026-09 – Present)

• Building hybrid speech emotion recognition models on IEMOCAP, including CNN-Transformer and spiking-neuron (LIF) architectures, as a modular PyTorch package, with cross-corpus testing planned
• Audited inherited data before training, recovering 501 empty recordings and regenerating spectrograms from raw audio to verify a collaborator's preprocessing; scripted multi-GPU sweeps on a 4x H100 node

Data Science Intern at Invent.us (2025-06 – 2025-11)

• Built an internal Python tool migrating Apache Superset dashboards between client environments in seconds, replacing per-client manual rebuilds for wealth management and VC clients
• Developed a 12-command CLI against the Superset REST API with per-tenant configuration, tag-based export, and automated import with mapping, iterated across releases in Git and Jira

E-commerce Operations Associate at GreaterGood.com (2024-05 – Present)

• Launched and scaled the company's Amazon marketplace channel from zero to $1M in annual revenue, running daily operations and building the catalog structure, listing standards, and processes from scratch
• Lifted conversion from 0.15% to a 2.4% trailing-year average through split testing on titles, images, and copy, using session and revenue-per-session data to decide which variants shipped
• Built the channel's reporting ecosystem in Excel and Google Sheets, merging Amazon settlement, ad, and returns feeds with internal cost data into monthly P&L, ASIN time series, and 20+ KPIs

Master of Science in Data Science – Gonzaga University (2027-05)

Bachelor of Business Administration in Business Analytics & Finance; Minor: Economics – Gonzaga University (2025-05)