DevOps Manager
Get-A-Head / Keel Digital
- Present
Leading a DevOps team managing a multi-account AWS platform for a healthcare SaaS company. I own all infrastructure decisions: Terraform architecture, container supply chain, CI/CD pipelines, and compliance posture across FedRAMP, HIPAA, and SOC 2. Grew from sole infrastructure hire to managing 4-7 engineers.
Key Achievements
- Led the team through FedRAMP authorization and got the product listed on the marketplace (FR2601536221)
- Moved 40+ containers from Docker Swarm to ECS Fargate with zero data loss. PostgreSQL replication bridge, secrets compatibility shim, and a coordinated cutover with rollback plan
- Set up a three-tier Terraform structure (67+ modules, templated stacks, per-account configs) managing 6 AWS accounts and 16 environments
- Put together the container supply chain security pipeline: FIPS 140-3 Chainguard images, cosign signing, Trivy + Grype scanning, CIS benchmarks, SBOM attestation
- Built the API gateway management system (Python, declarative YAML config for 40+ containers, 3 custom Lua plugins) and identity provider infrastructure (Keycloak FIPS builds, 15+ branded locales, CI/CD consolidation)
- Replaced 12+ per-brand Docker builds with a single-image architecture using Go CLI tooling. Build time from 50-60 min with an artifact per brand to a single 8-minute build
- Built a vulnerability consolidation tool in Python that pulls from GitHub, Vanta, and Jira into a single dashboard. 253 tests, 28 property-based
- Stood up the AI/ML infrastructure (Bedrock, SageMaker, pgvector) under the same Terraform and compliance controls as everything else
- Leading a distributed DevOps team of 4-7 engineers: run structured knowledge-sharing sessions, performance reviews, technical interviews, and 1:1s focused on individual growth. Wrote 2,500+ lines of multi-audience architecture documentation for a single project
- Manage the GitHub org, Google Workspace, DNS (20+ domains via OctoDNS), and coordinated releases across 20+ repos. Own the full release process from PR merge through production deployment