I build deterministic data and workflow systems for messy, exception-heavy operations, using Python, SQL, BigQuery, APIs, and Google Workspace.
My focus is practical automation: turning inconsistent inputs and operating rules into systems that can be inspected, reconciled, and approved by a person before consequential changes are accepted.
- Data ingestion and quality controls — structured pipelines that validate inconsistent files and operational feeds before downstream use.
- Reconciliation and exception workflows — controls that preserve missing or unpaid records, explain mismatches, and retain an audit trail.
- Warehouse-backed reporting — reusable SQL and bounded presentation layers that move business logic out of fragile spreadsheets.
- Product-data workflows — validation and approval patterns for structured catalog and operational data.
Python · SQL · BigQuery · pandas · REST APIs · Google Apps Script · Cloud Storage · n8n · data quality · reconciliation
I use AI coding agents to accelerate investigation, implementation, debugging, review, and documentation. I define the requirements, business rules, acceptance criteria, and safety boundaries; verify generated work against source evidence and deterministic checks; and retain responsibility for production approval. The systems themselves use deterministic runtime logic rather than AI decision-making.
Each repository below is an independently written clean-room reference: no employer or client source code, data, credentials, or identifiers. Each runs end to end from synthetic fixtures, with a deterministic demo and its own test suite.
- Commerce data automation platform — ingestion, governed warehouse layers, data quality, and thin reporting.
- Operations reconciliation controls — exception-preserving settlement and freight reconciliation patterns.
- Resumable Drive file pipeline — preview/apply safety, resume semantics, and duplicate-aware verification.