Rewiring the supply chain with intelligent data pipelines
I worked as the lead data engineer and developer on a major iteration of Phaxel's platform. We took B2B supply chain compliance in a bold new direction, focusing on becoming the best data ingestion tool for operations teams.
Visit websiteThe problem
In 2026, I began a project to build an entirely new platform from the ground up to serve as the most powerful tool for suppliers to manage messy B2B data exchanges. The industry standard relies on outdated EDI files, and there were a number of complex data engineering problems to solve in the process of automating pipeline validation. The primary goals for the project were reducing bottlenecks in data ingestion, and making the platform's outputs perfectly clean for retailer requirements, but with plenty of room to scale for predictive analytics.
The robust data architecture
To streamline the ingestion process across diverse suppliers and retailers for such a complex project, it was important to lay the foundations with a strong, flexible data pipeline that could evolve during the platform's development cycle. This would inform both the schema validation and historical error tracking across the backend itself as well as the client-facing UI and generated labels.
Schema validation & modeling
A data pipeline is useless if the outputs aren't standardized, so I put together a comprehensive validation engine to cover schemas, automated cleaning, type enforcement, and formatting guidelines for all incoming EDI payloads flowing through the system.
Data flow & transformation
Transformation was a core principle in making the ingestion experience a more reliable process. Payloads are intercepted and parsed in ways that highlight the cause and effect of each schema violation to improve the structural integrity of the overall database.
Encouraging data quality
A major part of solving for compliance was being able to visualize the payload errors in the UI. This was especially beneficial for operations managers and developers who need to review and apply automated fixes on the failing data structures without having to dig through all of the raw EDI text line by line.
An extensible validation engine
The most powerful aspect of the platform is the ability to apply custom validation rules for any retailer, whether it be a specific carrier code, missing field, or formatting error. Out of the box these checks run automatically with minimal compute overhead from the servers. Operations teams can then correct everything using a centralized React dashboard.
Next-generation data processing
The flexibility of the Python and PostgreSQL stack allowed the engine to process incredibly dense data formats as highly structured JSON that could then be manipulated by the frontend. Suppliers can load any raw text payload and instantly identify the exact lines triggering validation failures. Users are then directed to specific missing data points and shown suggested fixes, allowing them to auto-fill records and generate compliant shipping labels instantly. Furthermore, data engineers can query these historical error rates stored in the Supabase tables to predict future compliance bottlenecks without needing complex external tools.
YMoviesProject outcomes
Ultimately the project was highly successful in proving that complex B2B bottlenecks can be solved with intelligent data engineering, serving as a powerful portfolio centerpiece and a robust foundation for modern supply chain compliance.