Generate conformant data pipeline transforms
You want an AI agent to generate a batch of Python data-pipeline transforms that all must conform to your team's contract: each is a pure function taking and returning a typed DataFrame, fully idempotent, logging through the shared structured logger, and registered in a manifest. Your first vague prompt ('write a transform that dedupes user events') produced something that worked locally but mutated its input, logged with print(), and wasn't registered. Describe the CONTEXT and constraints you'd put in the spec so it gets it right the first time, and why each matters.
normalize_event_rows(rows: list[str]) → list[str][["u1|Click|3","U1|click|4","u2|view|1"]]out["u1|click|7","u2|view|1"][["u1|click|7","u2|view|1"]]out["u1|click|7","u2|view|1"][["a|x|600","a|x|700"]]out["a|x|1000"]Treat the AI’s output as a draft to verify, not an answer to trust. Name the specific flaw and the input that triggers it, say how you’d catch it (tests, edge cases, reading critically), and how you’d re-prompt or decompose to get it right.
Vibe & agentic: describe the solution in plain language (or narrate it) and the coach grades your approach.