Kevin Delgado

Day-to-day: untangling the hardest ML problems enterprise clients face — broken training pipelines, silent inference failures, data ingestion bottlenecks, and distributed computing edge cases. Deliver production-ready solutions across SageMaker, Bedrock, Rekognition, and Textract, wired into IAM, S3, Lambda, Redshift, Glue, DataZone, Snowflake, and Databricks. Beyond support, designed an internal PoC tool that improved troubleshooting efficiency for complex cloud architectures — adopted by 100+ AI/ML engineers worldwide across retail, healthcare, tech, and education.

Joined the consulting engineering team full-time to tackle large-scale data challenges. Orchestrated and optimized AWS Glue workflows for a pharmaceutical data migration affecting millions of records — cutting ETL latency by ~25% and significantly reducing error rates. In parallel, built and drove adoption of a Power BI dashboard that gave teams real-time visibility into worker productivity, delivering a 10–20% boost in performance tracking.

First production-grade engineering challenge: designed and deployed an end-to-end ETL pipeline measuring operational data across 8 manufacturing plants in the US and LATAM — boosting data throughput by 35% as measured by job completion time. On top of that, built a PoC decision-making dashboard for site leaders that consolidated plant-level insights into a single view, projected to reduce the decision-making cycle by ~40% upon full adoption.