AI/ML Architect and Staff Platform Engineer who owned Grindr's infrastructure through its NYSE IPO and 126% revenue growth — and now ships production-grade agentic systems faster than most teams ship features. I build cloud infrastructure, identity systems, and the automation that removes manual work from the loop.
Scale: 15M monthly active users, 190 countries, 8 AWS accounts, 5+ EKS clusters, $1.4M/year in cloud cost eliminated.
Infrastructure as code - Every system I own is Terraform-managed, PR-reviewed, and deployed without manual steps. If it isn't declarative, it isn't done.
Identity and access - Multi-account IAM federation, IRSA per-service scoping, OIDC-based CI/CD with zero static credentials. I've migrated orgs from legacy SSO to federated identity end-to-end.
Automation pipelines - Event-driven backends that propagate state changes across systems automatically. When something changes in one place, everything downstream reacts without a human in the loop.
Developer experience - CLIs, MCP servers, and environments that make engineers faster. If a person does the same thing twice, it gets automated.
Go devtools monorepo: thirteen CLI tools for shell optimization, structured telemetry, and AI-augmented workflows. Every tool emits to a unified JSONL event bus — usage-driven decisions about what to build, optimize, or deprecate. Includes mdq (markdown query engine), ts-go (tree-sitter Go analysis), perfgate, workctl, and more.
Statistical performance gating for CI. Define a benchmark, set a regression threshold, fail the build. Simple contract, serious signal.
JSONL event bus for AI-augmented engineering. Non-deterministic systems need observability that accounts for probabilistic outputs, not just pass/fail.
Terraform modules for GitHub + AWS + GCP identity federation from scratch. Zero static credentials from day one. Designed to be the first thing you run in a new org.
Declarative Terraform patterns for least-privilege IAM: role trust policies, boundary conditions, cross-account trust setups, and the guardrails that catch mistakes before production.
MCP server that surfaces AWS IAM and S3 introspection directly into AI coding assistants. Security context in the tool that is already open.
Farewell John (hersolutionsllc)
AI-assisted infrastructure for an early-stage startup: I own the stack from the public site through the API, data ingestion, deployment automation, and day-to-day operations. The design goal is simple — keep the request path fast, push crawling and parsing into the background, and make every change reproducible locally before it ships.
flowchart LR
Users --> CDN[CloudFront / demo]
CDN --> API[Read-only API]
API --> PG[(PostgreSQL + PostGIS)]
Crawlers --> AF[Airflow DAGs]
AF --> PG
AF --> Mail[Email / notifications]
CI[GitHub Actions + IAM] --> Infra[Terraform / AWS / GCP]
Infra --> API
Infra --> CDN
| Area | Outcome |
|---|---|
| Read path | Fast API responses from PostgreSQL only |
| Background work | Crawling, parsing, and ingestion moved off the request path |
| Local clone | One-command production-like environment for testing changes |
| Deployment | OIDC-backed CI/CD with no static credentials |
| Ops | Cost tracking, secrets handling, and workspace admin automated |
Go Python TypeScript HCL / Terraform AWS EKS ArgoCD Datadog PostgreSQL Airflow GitHub Actions
brianlopez.us · Chicago, IL