The Lab
Not everything I build starts as a production project.
The Lab is where I experiment with data, machine learning, AI and ideas I want to understand better. Some experiments become projects. Some fail. Some simply teach me something. Projects are work I've built and shipped: this page is things I'm exploring, nothing here is a finished, verified result.
Can we determine whether two marketplace listings represent the same underlying product?
ExploringGoal: build a system that can identify potentially identical or closely related products across different sources.
Can historical marketplace data help explain or predict product price behavior?
ExploringGoal: understand which variables contribute to price changes and whether useful predictive signals exist.
What factors are associated with better marketplace search visibility?
ExploringGoal: move from simply tracking ranking to understanding the factors associated with ranking performance.
Can language models improve extraction from difficult semi-structured data?
ExploringGoal: compare conventional extraction approaches with LLM-assisted workflows and understand where each performs best.
Can unusual marketplace behavior be detected automatically?
PlannedGoal: identify meaningful anomalies without overwhelming the user with false positives.
Something here worth discussing?
None of these are finished. I'm happy to talk through any of them as open questions.