Gustavo Marín's Work | ContraWork by Gustavo Marín
Gustavo Marín

Gustavo Marín

SQL, Python & Data Automation Specialist

Cover image for Migration Data Validation — SQL
Migration Data Validation — SQL & Python Demo This synthetic example shows why matching row counts and totals is not enough to validate a data migration. Both datasets contain 10 rows totaling USD55.00, yet row-level checks identify a missing record, a duplicate key, and a changed amount. Includes executable Python validation, sample datasets, a discrepancy report, and Snowflake SQL templates. Python checks were executed locally. The Snowflake SQL has not been executed against a warehouse. Demonstration using synthetic data, not client work.
1
5
Cover image for Power BI Sales Reporting —
Power BI Sales Reporting — Model & DAX Build Kit A reporting build kit using 720 synthetic sales orders, with date, product, and region dimensions. Includes CSV datasets, DAX measures, expected KPI values, and a Power BI Desktop build guide. The sample contains USD193,341 in revenue and a 41.93% gross margin. The preview was rendered in Python. No PBIX file is included; the DAX measures and report still require validation in Power BI Desktop. Demonstration using synthetic data, not client work.
0
8
Cover image for Python Reporting Automation — Synthetic
Python Reporting Automation — Synthetic Data Demo A reusable Python workflow that combines two CSV files, validates dates and amounts, handles duplicates, and produces clean data, a rejection log, a regional summary, and an audit report. The sample reconciles 25 input records into 20 accepted and 5 rejected records, with USD3,470 in accepted revenue. Includes source code, run instructions, and a validation test. Demonstration using synthetic data, not client work.
0
12
Cover image for SQL Query Optimization — Synthetic
SQL Query Optimization — Synthetic Data Demo This demo compares two SQL filtering approaches on 100,004 synthetic sales records using SQLite. Replacing a date-function filter with an indexed date range reduced median runtime from 20.511 ms to 3.967 ms across nine runs, while preserving identical results. The example includes query-plan analysis, result validation, and reproducible Python code. Demonstration using synthetic data, not client work. Performance varies by environment; these are local SQLite results, not Snowflake benchmarks.
1
18