Designed and developed scalable ETL/ELT pipelines integrating Amazon, Walmart, and TikTok APIs into AWS analytics environments, processing 4M+ records daily with data-quality validation and workflow documentation. Implemented a lakehouse architecture on AWS (Lambda, Glue, S3, Redshift) with structured transformation layers plus production pipeline monitoring and observability. Led the migration of legacy pipelines (Xplenty.io) to AWS, modernizing the ingestion architecture and reducing overnight processing windows from 6+ hours to under 45 minutes. Implemented Type 2 SCD patterns for churn analysis and data-processing pipelines for AI models (LangChain, FAISS, OpenAI GPT).

Built and optimized Snowflake analytics layers with dbt and Apache Airflow (100+ DAGs), consolidating CRM and operational sources into dimensional models that process 800K+ daily transactions. Designed data models and KPI dashboards in Power BI, translating business requirements for churn, LTV, and membership economics into repeatable metrics for non-technical stakeholders. Implemented data quality and integrity controls plus observability checks, reducing pipeline incidents from about three per day to zero and ensuring data availability and reliability. Automated document-classification workflows using LangChain, RAG, and OpenAI GPT APIs, reducing manual validation from about eight hours to under one hour per day.

Architected and implemented a medallion data platform (Bronze/Silver/Gold) in Azure Databricks and PySpark, processing 10M+ Telcel subscriber records with Azure Data Factory and Azure Event Hubs. Reduced data latency from 24 hours to under two hours and optimized Spark jobs from three hours to under 40 minutes through pipeline redesign and partition tuning. Applied Type 2 SCD patterns, idempotent pipelines, and data-governance standards; implemented CI/CD through Azure DevOps Pipelines and version control with Git. Mentored junior engineers in scalable design patterns, PEP 8 standards, and Git-based development workflows.

Developed ETL pipelines in Python and SQL to feed computer-vision models using 50K+ infrastructure inspection images, reducing preprocessing time from 6+ hours to under one hour. Built analytical datasets and anomaly-detection pipelines for infrastructure-damage detection workflows.