AI & Data Science in Fintech
Avatar of amelia-henderson amelia-henderson
amelia-henderson
August 8, 2025 at 9:02 am

What data do we actually need for ML underwriting?

What data do we actually need for ML underwriting?

I’m researching this for a project and would love real-world input. If you’ve built or shipped something here, what did you learn? Context: I’m Daniel Stewart and I’m evaluating options for our team.

4
4

  • Avatar of Alex Kugell

    Alex Kugell

    August 9, 2025 at 10:02 pm

    Biggest risk is invisible complexity creeping in. Keep the system operable by the smallest competent team. Ensure sandbox parity to avoid release‑day surprises.

  • Avatar of Marcelo Pantoja

    Marcelo Pantoja

    August 11, 2025 at 12:02 am

    Tune thresholds per segment; one size rarely fits all. Latency budgets forced us to drop nice-to-haves. Small anecdote that might help.

  • Avatar of Marcelo Pantoja

    Marcelo Pantoja

    August 12, 2025 at 6:02 am

    Keep dependencies boring and well‑documented. Keep the system operable by the smallest competent team. Biggest risk is invisible complexity creeping in.

  • Avatar of Daniel Alcanja

    Daniel Alcanja

    August 12, 2025 at 4:02 pm

    Quick checklist that saved us time. Ensure sandbox parity to avoid release‑day surprises. Run a pre-mortem; it surfaces surprises early.

Log in to reply.