What I use to ship
The stack behind the case studies — chosen because it's boring, reliable and fast to work with.
01 / AI
AI tooling
Claude & Claude CodeAgents, skills, MCP
OpenAI APITool use, structured output
LangChainAgent orchestration
MCP serversSecure tools for LLMs
Qdrant · pgvectorRAG retrieval
SageMaker · BedrockFine-tuning, hosting
AI SDKStreaming UIs
SupabasePostgres, auth, vectors
02 / Stack
Everything else
- Data
- PostgreSQL · MongoDB · MySQL · Redis · Prisma · Qdrant · pgvector · Pinecone
- Backend
- Node.js · NestJS · Express · GraphQL (Apollo) · REST · Socket.io · Microservices · Event-driven architecture
- Testing
- Jest · Mocha · Chai · Cypress · Contract & integration testing
- AI / LLM
- Anthropic API (Claude) · OpenAI API · LangChain agents · MCP servers · RAG · Embeddings · Fine-tuning · Tool use · Evals · Prompt engineering
- Frontend
- React · Next.js · Vue.js · Nuxt.js · Angular · Redux · Tailwind CSS
- Languages
- TypeScript · JavaScript (ES2022+) · Python · SQL
- Cloud & DevOps
- AWS (EKS, ECS, Lambda, S3, SageMaker, Bedrock) · Docker · Kubernetes · Jenkins · GitHub Actions
- Fintech & payments
- E-wallet transaction systems · Audit trails · Reconciliation · Stripe · PayPal · Paymob · Paytabs · OAuth 2.0 · RBAC