4+ years designing enterprise-scale data pipelines and production ML systems — from Snowflake ML model registries to Azure DevOps CI/CD — for global CPG clients across sustainability, procurement, and traceability.
I design MLOps pipelines on Snowflake ML and Snowpark — metadata-driven feature engineering, model training, registry, versioning, and automated batch prediction — orchestrated through Azure DevOps CI/CD. Deep hands-on work across Snowflake, Azure, Databricks, PySpark, and dbt Core, delivering anomaly detection and monitoring for global CPG clients.
A production-grade MLOps pipeline that ingests emission activity data, engineers features via a metadata-driven layer, trains and registers models in the Snowflake Model Registry, and serves automated batch predictions — fully automated through Azure DevOps CI/CD.
Extended to detect anomalies at multiple grains (SKU, zone, category) without duplicating code, surfaced through Streamlit and Power BI.
A natural-language querying interface over the GPS Ingredients data model using Snowflake Cortex Analyst, letting non-technical stakeholders self-serve emissions insights without writing SQL.
Defined a YAML-based semantic layer mapping 15+ dbt models to business-friendly metrics, cutting estimated ad-hoc reporting requests by 40%.
Open to conversations about MLOps, Snowflake ML, and enterprise-scale data architecture.