Duration: ≈1 week.
You already have what most AI learners spend months acquiring: REST/API design, Postgres, error handling and retries, queues (SQS/n8n), Docker, AWS, deployment. Don’t relearn these — reuse them.
The one gap to close
Enough Python to read notebooks and use Python-only tools — the ecosystem leans Python. But you can build production AI in your existing TS/Node stack: the Vercel AI SDK is excellent and JS-native.
Don’t gate progress on “mastering Python first”
Pick Python up as a second language alongside building. You only need to read notebooks and run Python tools — not become a Python expert before you start.What you’re carrying forward
| You already own | Where it pays off in AI |
|---|---|
| REST / API design | LLM wrapper services, tool endpoints |
| Postgres | pgvector for RAG — zero new infra |
| Retries & error handling | Reliable LLM calls (they fail constantly) |
| Queues (SQS / n8n) | Async/background AI jobs, batching |
| Docker / AWS | Deploying AI services & self-hosting tools |
Milestone
You can read a Python notebook, run Python-based AI tools, and you’ve decided your default build stack (TS/Node + Vercel AI SDK) without blocking on Python mastery.