Approach
How I work through a problem.
The methods I use and the tools I rely on, in roughly the order I reach for them.
Toolkit
Programming
- Python
- C++
- SQL
- TypeScript
- MATLAB
Data & BI
- Tableau
- Power BI
- DBeaver
- pandas
- NumPy
AI & ML
- YOLOv8
- scikit-learn
- PyTorch
- Edge AI
- Smart sensors
Ways of working
- Agile (Scrum)
- Project management
- Stakeholder communication
- Analytical writing
Method
01 · Define the decision
Begin with the decision the analysis must support. Without a decision in mind, there is no analysis worth running.
02 · Profile the data
Examine shape, volume, freshness and gaps. Understand what the data is before deciding what it should say.
03 · Run the smallest useful experiment
Choose the cheapest test that could disprove the hypothesis. Speed of feedback compounds faster than cleverness.
04 · Hand it off cleanly
Document assumptions, write SQL that reads naturally, and leave tests behind. The next maintainer is as much a stakeholder as the first reader.
05 · Close the loop
Report what was decided, not only what was learned. Verify whether the decision held and update the model of the world accordingly.