Data Engineering II — Python
Over 6 years, Braden Keith grew a minimal no-code Domo solution into a proprietary Python SDK microservices platform to support Global Ordnance's multi-channel growth and process millions of records per day with high efficiency and stability.
Overview
Name
Braden Keith
Title
Data Engineer
Employer
Global Ordnance LLC (GO)
Sector
eCommerce
Industry
Commercial Defense
Employment
2019 – 2025
Via
Upwork · Contract-to-Hire
Contract
"Domo help"
"Absolutely fantastic, knowledgeable and incredible guy."
About
Braden Keith joined Global Ordnance as a freelance contractor in 2018, shortly after his tenure at Sandia National Laboratories, where his work as a Business Analyst had propelled his early career. In 2019, he secured a contract with Global Ordnance under the project codename "Domo help", tasked with automating operations to manage the Extract, Transform, Load (ETL) of Inventory and Pricing on GO's BigCommerce-hosted eCommerce site. Using Domo's no-code Graphical User Interface (GUI), Braden engineered initial solutions that streamlined daily inventory and pricing updates without writing a single line of code.
From July 2019 to May 2025, as GO's Data Engineer, Braden Keith progressively modernized the company's data pipeline infrastructure. Early in this role, he leveraged Domo's drag-and-drop ETL workflows for quick automation wins, rapidly integrating multiple vendor data sources and eliminating manual data handling in the eCommerce platform. As GO's business expanded and data requirements grew more complex, Braden transitioned these processes to a robust Python-based microservices architecture. He developed a proprietary Software Development Kit (SDK) to enable seamless integration between systems, and deployed a network of lightweight microservices to handle data extraction, transformation, and loading at scale. This evolution from a no-code tool to a custom-coded platform not only increased flexibility but also improved performance and reliability of the pipeline.
Objective
Optimize Vendor and BigCommerce integrations with Domo.
Key Challenges
API Diversity
Each endpoint had unique rate limits, pagination strategies, authentication methods (OAuth2, API key, token-based), and schema structures.
Data Mapping
Schema mismatch required flexible data transformations and field mapping logic that could be reconfigured per integration.
Error Handling
Needed to gracefully manage transport failures, timeouts, and data validation errors without halting the pipeline.
Monitoring
Required a logging and notification system for transparency and debugging.
Reusability
The solution had to be generalizable, enabling new integrations to be configured — not coded — through a web UI or YAML/JSON config.
Architecture & Methodology
Proprietary Python SDK
The foundation of the orchestration layer was a custom-built Python SDK designed to abstract and manage API Client Adapters (each third-party API modeled using pluggable client adapters wrapping requests in unified methods for authentication, rate-limiting, and pagination), Data Transformers (custom Python logic to handle schema mapping, validation, and payload transformation), and Dataflow Templates (reusable job templates describing data sources, targets, transformation rules, and scheduling).
Web Application
The SDK was surfaced through a lightweight Flask web interface that allowed non-technical users to configure API endpoints and credentials securely, and monitor pipeline status and logs in real-time.
Orchestration Layer (Job Runner)
Jobs were queued and executed asynchronously using AWS Lambda, ensuring scalability across large data volumes, retry handling and graceful degradation, and parallel execution of independent flows.
Observability & Logging
Integrated with AWS CloudWatch and local logging for job status monitoring, detailed error logs, and alerting on failed runs.
Tech Stack
Language
Python 3
Framework
Flask (web UI), AWS Lambda (job runner)
Cloud
AWS Lambda, AWS CloudWatch, S3
Integrations
BigCommerce API, Domo API, 20+ vendor APIs
Auth Patterns
OAuth2, API key, token-based
Config
YAML / JSON dataflow templates
Observability
CloudWatch Logs, custom alerting
Prior Foundation
Domo (Magic ETL → MySQL DataFlows)
Skills Demonstrated
Key Outcomes
Lower Latency
Data refresh times were slashed from hours to near real-time, ensuring that online inventory levels and prices were always up-to-date across all channels.
Improved Accuracy
Real-time synchronization eliminated stock discrepancies, providing a single source of truth for inventory and guaranteeing accurate stock information for both customers and internal teams.
Dynamic Pricing
Live data integration enabled responsive pricing strategies, allowing GO to adjust product prices in real time based on demand, competitor prices, and available inventory.
Higher Uptime
The microservices orchestration introduced isolated failure domains – if one service went down, it no longer brought the whole system down – thereby increasing overall system uptime and reliability.
Outcome & Impact
Faster Onboarding
Reduced the time to onboard a new API from weeks of custom code to minutes of configuration.
Reliable Sync
Maintained real-time or scheduled dataflows across multiple platforms with over 99.99% uptime.
Scalability
Currently handles millions of records across dozens of jobs daily with minimal manual intervention.
Continuation of
Data Engineering I — MySQL
Braden Keith + MySQL = System Optimization