I architected the data and AI infrastructure for a sustainability platform that automatically estimates product carbon footprints. The platform processes diverse data sources — including product specifications, manufacturing data, and supply chain information — to generate accurate emissions estimates at scale.

The architecture encompasses three main layers: a data ingestion and orchestration layer that normalizes heterogeneous input data, an AI services layer that applies estimation models to structured product data, and an analytics layer that surfaces insights to enterprise customers through dashboards and APIs.

This work was undertaken during a critical infrastructure recovery and modernization phase, where the goal was to establish scalable, reliable foundations that would support the platform's growth and enable intelligent automation of sustainability workflows.

The data pipeline layer was built with automated ETL workflows designed to handle the heterogeneity of sustainability data. Sources range from structured product databases to semi-structured supplier reports, requiring flexible parsing, validation, and transformation logic with robust error handling for data quality issues.

The AI services component implements carbon estimation models that take structured product and supply chain data as input and produce emissions estimates. These services were designed as modular, independently deployable units with clear API contracts, enabling the team to iterate on estimation accuracy without disrupting the broader platform.

The entire stack was deployed on AWS with comprehensive monitoring, alerting, and cost optimization. Infrastructure-as-code practices ensured reproducible deployments, while observability tooling provided visibility into pipeline health, model performance, and system reliability in production.

The platform enabled enterprise customers to access real-time sustainability insights, replacing manual and time-consuming carbon estimation processes with automated, scalable analytics. This shift allowed sustainability teams to make data-driven decisions about product design and supply chain optimization.

The scalable infrastructure foundations established during this phase positioned the platform for growth, with the architecture designed to accommodate increasing data volumes, additional data sources, and more sophisticated estimation models without requiring fundamental redesign.

The modernization effort successfully recovered and improved the platform's core capabilities, establishing reliable CI/CD pipelines, comprehensive monitoring, and production-grade operational practices that reduced incident response times and improved overall system availability.