The kind of problems we like to solve.

Challenge

Finance, merchandising, and marketing each had their own version of "revenue," and closing the books each month meant days of manual reconciliation across POS, e-commerce, and ad platforms.

Approach

• ▸Stood up a cloud warehouse and ingestion pipelines from POS, Shopify, and ad platform APIs
• ▸Built a dbt transformation layer with tested, documented metric definitions
• ▸Delivered exec, merchandising, and marketing dashboards on a shared semantic layer

Challenge

Retention efforts were reactive: by the time churn showed up in monthly metrics, the behavioral signals that predicted it were weeks old.

Approach

• ▸Engineered behavioral & usage features from product event data
• ▸Trained and evaluated a churn-risk model, benchmarked against a naive baseline
• ▸Shipped risk scores into the CRM to trigger early retention workflows

Challenge

Knowledge existed but wasn't searchable in one place, and generic chat tools couldn't answer questions accurately without hallucinating.

Approach

• ▸Built a retrieval-augmented generation (RAG) pipeline over internal docs, wikis, and PDFs
• ▸Added citation-backed answers so every response linked back to its source
• ▸Deployed as an internal chat tool with usage monitoring and feedback loops