BKV Pulumi Templates - BKV Corporation

Developer: Manuel Solalinde (manolosolalinde@gmail.com)

An internal infrastructure-as-code library that standardizes how the data and engineering teams ship containerized applications to AWS. Instead of each project re-implementing networking, authentication, load balancing and CI/CD from scratch, every app starts from a small set of reusable building blocks and ready-made scaffolding — so a new app is deployable in minutes rather than rebuilt from the ground up. It eliminates per-project re-implementation of the same infrastructure and is the deployment foundation behind 6+ production apps: Surveillance Dashboard, Gas Targets, Gas Injection, Plunger Optimization, AI Agent Chat, and the broader Streamlit toolkit.

What it provides

• A single application-stack component — one declaration provisions a complete, production-ready ECS Fargate application end to end: container registry (ECR) with Docker build and push, an Application Load Balancer fronted by corporate SSO (Azure AD OIDC), DNS and TLS (Route53 + ACM), WAF, scoped IAM, centralized logging (CloudWatch), and target-tracking autoscaling. The whole library is implemented as reusable Pulumi ComponentResources, so apps assemble infrastructure from typed, parameterized parts rather than hand-wiring resources.
• Project scaffolding — generates a new, immediately deployable project pre-wired to the library and to CI/CD: GitHub Actions run pulumi preview on every pull request and pulumi up on merge, so every scaffolded app ships with preview-on-PR and deploy-on-merge from day one.

Design decisions & trade-offs

• One orchestrator over many loose components — a single top-level application-stack component composes the underlying pieces instead of asking each project to wire them together by hand. This reduces parameter explosion and enforces consistency across apps.
• Composable add-ons over a monolithic template — analytics integrations attach to an existing application stack rather than baking every integration into one template. Teams adopt features gradually, without rewriting the base deployment.
• Consume shared infrastructure, don’t redeclare it — applications consume the shared base network (VPC, subnets, ECS cluster, WAF) through a stack-reference pattern instead of re-creating it. One owner manages the base network; apps simply reference it.
• Dual-path corporate auth — SSO at the load balancer (Azure AD OIDC) governs app access, while SAML federation governs analytics query access. Keeping these as separate identities lets app sign-in and data-query permissions evolve independently.

Hard problems solved

• Standardized enterprise authentication for every app — corporate SSO (Azure AD OIDC) at the ingress plus SAML-federated analytics access, so every app gets enterprise sign-in essentially for free instead of reinventing auth each time.
• Least-privilege cross-account data access — secure, scoped access into a central Redshift data warehouse living in a separate account, granted through an assumable role rather than broad, shared credentials.
• Reusable multi-environment deployments — the same code yields dev, stage, and prod environments, with secrets sourced at runtime from Pulumi secrets and AWS Secrets Manager and never placed in the codebase.
• Consistent tagging and naming at scale — a uniform, predictable scheme for naming and tagging every resource so cost allocation and ownership tracking stay reliable across a growing fleet of apps.

Key Features

• One-command application stack — a single application-stack component provisions ECR, ALB, DNS, certificates, WAF, IAM, logging, the Fargate service, and autoscaling.
• Corporate SSO built in — Azure AD OIDC at the load balancer for app access, and SAML federation for analytics query access.
• Reusable and multi-environment — components are parameterized and environment-scoped, so the same code deploys dev, stage, and prod.
• Standardized CI/CD — scaffolded projects ship with GitHub Actions preview-on-PR and deploy-on-merge workflows.
• Secrets-safe by design — secret values flow through Pulumi secrets and AWS Secrets Manager, never the codebase.
• Composable add-ons — Athena and Redshift integrations attach to an existing application stack without rewriting it.