Tim Radcliffe
Data Science & Experimentation
I'm a data scientist who's spent the last few years thinking about why experimentation programs succeed or fail—both the statistical and human parts.
~13 years building software, data products and teams.
Recent Projects
→ Tyche
Browser-native Bayesian A/B testing. Answers "what's the probable effect?" not "is p < 0.05?" Everything runs client-side—your data never leaves your machine.
→ Why Your Data Team Keeps Quitting
Data scientists average 1.7 years per role. 70% of data engineers are actively job hunting. The problem isn't compensation—it's how we structure teams.
→ Data Governance Without the Theater
Data governance has a reputation problem. Here's what actually works: contracts in code, docs that live with queries, and regular "what broke and why" sessions.
→ AI Agents: Infinite Queue of Fresh CS Grads
They have eidetic recall of Stack Overflow but zero context about your system. Here's how to use compile-time controls to prevent them from architecting your CRUD app like Google.
Currently
Available for consulting and full-time opportunities where experimentation and causal inference matter. Based in Vancouver, comfortable with remote.
I'm most excited about long-term relationships with companies building genuinely good products. The sweet spot: problems at the intersection of development, business understanding, and modern statistics—where we're embedding data-driven decision making deeply into organizational structure, not just dashboards. Particularly drawn to organizations that blend quantitative rigor with qualitative insight, and those working to improve well-being, protect wild places, or advance privacy and autonomy. More about my approach →