Matthew Nelson

Architected and implemented production-grade Dagster/dbt pipeline processing USA oil & gas data across 15+ states with 400+ transformation models, replacing legacy Python-based ETL processes with maintainable modern data stack. All Open Source & Self-Hosted.

Reduced processing time by up to 98% for critical datasets through strategic use of Polars, DuckDB, and optimized data modeling patterns

Designed and deployed AWS cloud infrastructure using Terraform, managing RDS PostgreSQL, ECS compute clusters, and S3 data lakes across staging and production environments

Implemented comprehensive CI/CD workflows with GitHub Actions, automated validation (formatting, linting, type checking), and multi-environment deployment strategies

Developed reusable framework patterns to standardize data pipeline development, reducing boilerplate code and ensuring consistency across 100+ data assets

Established software development best practices including schema contracts, comprehensive testing, type-safe code, detailed documentation, and collaborative Git workflows for enterprise-grade reliability

Created a multi-level Netback & Operating Cost Dashboard in PowerBI. Planning and forecasting functionality was embedded in a beautiful and intuitive interface

Designed a Corporate Capital Dashboard in PowerBI, with a similar multi-level structure to the Operating Cost Dashboard and consistent design styling.

Redesigned a Weekly Well Production report which was generated manually in Spotfire weekly. The new report implemented best practices in data visualization to greatly improve readability, and the full update/pdf/email process was automated using Spotfire's Automation Services and Microsoft's Power Automate.

Built a simple Lift Gas Optimization Dashboard in PowerBI to help the Production Engineers understand the lift gas usage across the company, and to identify opportunities for optimization.

Replaced a completely manual task of choosing the sequence for a zipper fracturing design on new well pads with a Python web app. The app was built with Streamlit, and integrated a purpose built algorithm to intelligently choose the next stage to stimulate while reducing operations time at the surface and staying 'ahead' of offsetting wells on frac azimuth.

Designed a Streamlit app to provide an interface for estimating the level of development required to fill a gas processing plant to capacity, taking in a base forecast, type curve forecast(s), and simulation parameters specific to the situation.

Fully implemented a local, modern, end-to-end data stack with multiple existing data processes replicated using python (ingestion), dbt core, cube.js semantic layer, DuckDB and PowerBI. Replaced antiquated processes using R and serving data as csv with a single executable which processes data from multiple databases, orchestrates an ELT DAG, and serves clean data in a semantic layer to PowerBI. Benefits from all of the features of dbt and cube.js. Can be spun up on any machine as a multi-service architecture with docker.

As a senior resource with significant cloud and analytic stack experience, advocated for the use of a true cloud based data analytics stack. Educated on the importance of thoughtful design for data ingestion, data engineering, semantic layers, and data visualization / data consumption.