David

Challenge: Finance teams spent hours every cycle pulling evidence, reconciling data and drafting documentation before reviewers could test recurring controls.

What I did: Built AI-assisted workflows that handle the prep work: gathering source data, flagging exceptions and drafting first-pass documentation for reviewers. The design keeps people in charge. AI prepares the work, and control owners review and sign off.

Result: Big cuts in manual prep time, more consistent documentation, and reviewers spending their time on judgment instead of assembly.

Tools: LLMs (Claude, ChatGPT), SQL, Python, prompt engineering

Challenge: Finance teams defined the same metrics in different ways, so dashboards, reports and new AI tools gave conflicting answers.

What I did: Designed a semantic layer across finance functions with governed metric definitions, documented business logic and reusable dbt models that people and AI tools both query.

Result: One source of truth for finance metrics, and a foundation that lets AI tools answer questions from trusted numbers instead of raw tables.

Tools: dbt, SQL, semantic layer design, data governance

Challenge: A month-end close process took about two weeks of manual work across Finance and the teams that depended on it, which left outstanding payments unclear until it finished.

What I did: Mapped the full process, automated the data pulls, reconciliations and reporting end to end, and built visibility into outstanding payments.

Result: The process went from about two weeks to about two hours, and Finance got same-day visibility into outstanding payments.

Tools: SQL, Python, dbt, workflow automation