juancompte

• Developed a personal project analyzing wallet behavior around ETH price drops (>10%) anomaly detection logic. • Built a modular DAG-based pipeline using GCP , BigQuery, DuckDB, and Dagster; focused on ETL optimization and exploratory analysis. • Engineered time series vectors per wallet and applied unsupervised clustering (K-Medoids with DTW) to uncover behavioral patterns, anchored to key price events for anomaly detection and trend analysis. • Stack: Python, GCP , BigQuery, DuckDB, S3, Dagster

Redesigned stakeholder-facing executive dashboard with business-critical KPIs, alerting logic, and a simplified layout — enabled leadership to stop mortgage leads based on poor conversion quality and bot traffic during quarterly reviews. Collaborated cross-functionally to launch Twilio-based proxy tracking and PII-safe reporting — improved client traceability, enabled compliance, and doubled call center revenue within 6 months through a new B2B service line. Created a consultative reporting solution to map ad spend across the funnel and resolve attribution issues — corrected Meta timezone mismatch, fixed ad group ID/session category logic, and avoided $2,000/day in suspended campaigns. Designed experimentation workflows using SQL and Python; translated results into actionable insight stories by funnel stage and customer segment — drove alignment on strategy with Marketing and Executive teams. Built a Unit Economics dashboard linking web sessions, paid ad data, and conversions — resolved cost misattribution, improved campaign ROI tracking, and increased marketing trust. Developed and owned the end-to-end billing automation by ingesting Salesforce CRM data into Redshift, creating warehouse snapshots, and integrating finance dashboards — reduced monthly effort from 6 to 2 hours and improved stakeholder satisfaction.

Developed dashboards to track cashier anomalies and branch-level discrepancies across 500+ Latin American branches. Improved predictive models for cash logistics by integrating Qlik Sense with Oracle-based data pipelines — reduced error margin by 18%. Led a data reliability project to reconcile ETL inconsistencies between retail and central operations, improving trust in reporting by 30%.

Created real estate & urban development dashboards — increased transparency and reporting speed by 35%. Audited KPIs, validated pipelines, and created documentation to ensure alignment with fiscal reporting requirements.