Ramanjit Singh Available for AI projects

The stack

Technologies I work with

Tools follow the problem, not the other way around — these are the tools I’m fastest with. Every item below appears in the six case studies — in the shipped demo or as the documented production path.

Plugs into what you already run

The center is the part I build

Swipe sideways for the full diagram →

Website · CRM · calendars · sheets · phones stay yours — the AI layer wires in around them.

AI models & inference

The reasoning layer

  • OpenAI (GPT family)Generation · Tool use
  • Anthropic ClaudeAgents · Long context
  • LLaMA — self-hostedPrivate deployments
  • Embedding models (e5, bge)Semantic representations
  • Speech pipelines (ASR / TTS)Voice agents

Retrieval & data

Grounded answers, verifiable sources

  • PostgreSQLStructured data · SQL answers
  • Vector stores (pgvector)Similarity search
  • RAG pipelinesChunking · Citations
  • OCR extractionScans → searchable text
  • Semantic searchMeaning over keywords

Automation & integrations

The busywork layer

  • n8nSelf-hosted orchestration
  • Make · ZapierManaged workflows
  • Webhooks & REST APIsSystem-to-system glue
  • Google Workspace APIsSheets · Calendar · Gmail
  • SMS gatewaysFollow-ups · Confirmations
  • Canva APIAsset generation in-flow

Engineering

How the products get built

  • Python · FastAPIBackends · AI services
  • Node.js · TypeScriptAPIs · Tooling
  • ReactDashboards · Product UI
  • WebRTC · WebSocketsReal-time voice & data
  • Tailwind CSSDesign systems in code

Infrastructure & delivery

Shipping and running it

  • DockerReproducible deployments
  • Cloudflare WorkersEdge hosting · This site
  • Self-hosted LLM infraData stays yours
  • GitHubVersioning · CI · Handover