Engineering Documentation
The technical foundation that powers MarSci in production.
Architecture, services, data, models, and infrastructure — designed for scale,
built for reliability, documented for clarity.
Documentation
The full stack, documented
React + Vite frontend and FastAPI backend. Auth flows, CSV upload, KPI computation, and the MMM configuration UI.
Airbyte for structured sources, Airflow for complex DAGs, SQS for async ingestion and job dispatch.
Causal-Driven Attribution and Bayesian Marketing Mix Modeling with Google Meridian. Results stored in DynamoDB.
GitHub Actions pipelines for build, test, and deploy. Push to main goes live in ~2 min via S3 and CloudFront.
EKS with Karpenter autoscaling, ALB controller, and Terraform for all AWS resource provisioning.
AI agent layer for autonomous marketing analysis, report generation, and insight surfacing across the MarSci stack.
Cross-platform Electron app for offline access, local data processing, and native OS integrations.
Academic and applied research underpinning MarSci's measurement methodology — causal inference, Bayesian statistics, and marketing science.
YouTube
See MarSci in action
Foundation