Projects
These projects cover two sides of my work: understanding how information behaves in search and generative systems, and investigating structured data with reproducible analytical methods.
Reading guide
These are independent portfolio projects, not client engagements. A research protocol, working analysis pipeline or proposed optimization is not evidence of improved search visibility or commercial performance. Each case study identifies its current status and distinguishes source material, analysis and measured outcomes.
AI Search · GEO · AEO
A reproducible framework for controlled query sets, brand visibility measurement, citation and retrieval analysis, automated reporting and testing.
A real public-site research subject taken from baseline through diagnosis, query fan-out, entity analysis and proposed optimization.
A controlled comparison of Notion, Asana, ClickUp and monday.com designed to keep first-party evidence separate from AI observations.
Data Analytics
Nine related datasets audited with Python/pandas, loaded into a Snowflake RAW layer and developed through dbt staging and intermediate models.
Analysis of portfolio exposure, claim severity, concentration and data quality, with an executive Tableau reporting layer.
Method
Where a project uses a benchmark, the prompts and scoring rules are fixed before results are interpreted. Where it uses public evidence, observations and inference stay separate. Where it uses data, transformations and quality checks are documented rather than hidden behind a final chart.