Platforms & Stack
n8n is one tool I build with. The architecture is what I'm actually selling.
Every case study on this site happens to run on n8n because it was the right call for that build — fast to ship, transparent, no vendor lock-in. The same discipline (validation, confidence routing, error handling, audit trails) travels cleanly to whatever platform your team already runs on, or to a fully custom backend when no-code tooling won't cut it.
Orchestration
n8n · Make · Zapier
Visual workflow platforms for fast, maintainable automation — picked per client based on budget, team familiarity, and hosting requirements.
Custom & SaaS
Python · Node.js · APIs
When a workflow tool hits its ceiling — heavy compute, custom auth, a real product to ship — I build the service directly and expose it as an API or a full SaaS backend.
AI Engine
Claude (Anthropic) · OpenAI · Gemini
Structured-output LLM calls, RAG pipelines, and multi-step agents built with LangChain or native agent SDKs — schema-validated, never hoped-for. Model Context Protocol (MCP) for wiring agents into real tools and data sources, not just chat.
Data & Vector Search
PostgreSQL · PGVector · Pinecone · Redis
Relational data where it belongs, vector search where semantic retrieval matters, cache where speed matters.
Infrastructure
AWS · GCP · Docker · Oracle Cloud
Self-hosted or managed, containerized where it helps, sized to what the workload — not a vendor's pricing tier — actually needs.
Integrations
HubSpot · Stripe · Slack · Google Workspace
The tools your team already lives in — wired in, not worked around.
