Data scientist · Founder · Researcher & teacher
I find the signal, build the business, and teach what I learn.
Biochemist turned data scientist. I build interpretable machine learning for health, founded and engineered BitSize Learning, and have spent a decade teaching people to think with data.
The practice
Three sides, one way of thinking.
Each side feeds the others: research sharpens the models, products force them to be useful, and teaching makes them clear.
01Data science
Data Scientist
Interpretable machine learning for health, from 400K-feature methylation arrays to 140K-record clinical EHRs.
97.4%held-out glioma accuracy
93–100%external validation
140KEHR records modeled
02Ventures
Business Builder
I spot a problem, design the product and write every line of it, then run it. BitSize Learning is the flagship.
1BC company founded
3platforms built & run
22learning apps shipped
03Research & teaching
Researcher & Teacher
Papers on genomics and the gut microbiome, peer reviewing for journals, and a decade of teaching data, code and biology.
7research papers
29K+publication views
9 yrsmentoring scientists
Now
On the bench right now.
What I’m running and reviewing this season, beyond the flagship and the paper below.
Reviewed
Peer review
BMC Microbiology
Reviewed a study of gut microbiota and host metabolism, accepted in August 2026.
About my reviewing
Pilot
Platform
Mr Pretzels Hub
Scheduling, BC payroll, daily sales and team chat, live with a real crew.
See the platform
Operating
Client build
AG3L website
Built and monitored the brokerage site, plus the carrier-data integration behind it.
Details
Flagship venture
BitSize Learning
A company I registered in British Columbia and a platform I designed and engineered myself. One Django codebase runs three sites (the hub, BitSize Python and BitSize Languages) with short, focused lessons in Japanese, French and Python.
3subscription tracks
22learning apps
5,200+lessons
From the lab
All case studies →
Selected data science.
97.4% · Held-out subtype accuracy
Interactive charts
Glioma Subtype Classification & Survival Modeling from DNA Methylation
Classifying brain-tumor subtypes and predicting survival from DNA methylation.
Read the case study →
140K+ · EHR encounters processed
Interactive charts
Predicting PHQ-9 Depression Scores from Multi-Clinic EHR Data
Predicting depression severity from routine EHR data.
Read the case study →
4 archs · Apples-to-apples benchmark
Comparative Evaluation of Pretrained Models for Lung Disease Classification
Benchmarking transfer-learning architectures on chest X-ray classification
Read the case study →
Research & teaching
Curiosity, shared out loud.
In the classroom
How I teach →
From lecture halls to 60-second lessons.
37MDS students taught Python & R
30+researchers through workshops
11STEM courses as TA
9 yrsmentoring grad students