The Stack

Tools I useto ship production.

Every layer of this stack runs in production today — across cloud infrastructure, AI systems, billing, and observability. No speculative tooling. No frameworks I haven't deployed.

Cloud & Infrastructure

AWS
Cloud platform
Terraform
IaC, single source of truth
AWS Lambda
Serverless compute
API Gateway
REST + WebSocket edge
CloudFront
CDN + edge caching
ECR
Container registry
Docker
Lambda container images
IAM
Identity & access boundaries
S3
Object storage

Application Layer

Next.js 16
App Router, RSC, edge
React 19
UI primitives
TypeScript
Strict mode, no any
Tailwind v4
Utility-first styling
motion/react
Animation system
Vercel
Frontend hosting

Backend & Services

Python 3.11
Service language
FastAPI
Async REST framework
Mangum
ASGI to Lambda adapter
Node.js 20
Lambda runtime + tooling
ThreadPoolExecutor
Regional fan-out scanning

AI Systems

Claude (Anthropic)
Primary LLM
AWS Bedrock
Managed inference
Anthropic SDK
Direct API + streaming
OpenAI SDK
Comparison + fallback
Google ML Kit
On-device pose detection

Data & State

DynamoDB
NoSQL primary
PostgreSQL
Relational, via Supabase
Supabase
Auth + Postgres + realtime
AWS Glue
ETL + crawlers
Amazon Athena
Serverless SQL on S3
CUR 2.0
Cost & usage data lake

Identity & Commerce

AWS Cognito
User pools + Google IdP
Clerk
Drop-in auth
Lemon Squeezy
Subscriptions + global tax
RevenueCat
Mobile IAP

Observability & DevOps

Sentry
Error tracking + perf
PostHog
Product analytics
CloudWatch
AWS-native logs + metrics
GitHub Actions
CI/CD pipelines

- AWS
  Cloud platform
- Terraform
  IaC, single source of truth
- AWS Lambda
  Serverless compute
- API Gateway
  REST + WebSocket edge
- CloudFront
  CDN + edge caching
- ECR
  Container registry
- Docker
  Lambda container images
- IAM
  Identity & access boundaries
- S3
  Object storage
- Next.js 16
  App Router, RSC, edge
- React 19
  UI primitives
- TypeScript
  Strict mode, no any
- Tailwind v4
  Utility-first styling
- motion/react
  Animation system
- Vercel
  Frontend hosting
- Python 3.11
  Service language
- FastAPI
  Async REST framework
- Mangum
  ASGI to Lambda adapter
- Node.js 20
  Lambda runtime + tooling
- ThreadPoolExecutor
  Regional fan-out scanning
- Claude (Anthropic)
  Primary LLM
- AWS Bedrock
  Managed inference
- Anthropic SDK
  Direct API + streaming
- OpenAI SDK
  Comparison + fallback
- Google ML Kit
  On-device pose detection
- DynamoDB
  NoSQL primary
- PostgreSQL
  Relational, via Supabase
- Supabase
  Auth + Postgres + realtime
- AWS Glue
  ETL + crawlers
- Amazon Athena
  Serverless SQL on S3
- CUR 2.0
  Cost & usage data lake
- AWS Cognito
  User pools + Google IdP
- Clerk
  Drop-in auth
- Lemon Squeezy
  Subscriptions + global tax
- RevenueCat
  Mobile IAP
- Sentry
  Error tracking + perf
- PostHog
  Product analytics
- CloudWatch
  AWS-native logs + metrics
- GitHub Actions
  CI/CD pipelines

Certifications & Objectives

Production-grade work first, paper second. These exams are on the roadmap to formalise what's already shipping in production.

AWS Certified Solutions Architect

Associate

Target: Q3 2026

AWS Certified DevOps Engineer

Professional

Target: Q1 2027

The list grows when there's a real problem to solve. It shrinks when something stops earning its complexity budget.