Skip to main content

Apply new skills to your role

Instruction and application
In Progress

In this hackathon workshop, you focused on packaging, deploying, and version-controlling an ML solution — and wiring in the explainability foundations that make it trustworthy in production. The same approach can be applied directly in day-to-day work.

Ship a reproducible first version​

Use the same mindset from the hackathon when approaching new problems:

  • Aim for a small, reproducible deployment before optimising for scale.
  • Version everything that can drift — code, config, model artefact, data schema.
  • Design rollback into the first release, not after the first incident.

Make decisions explainable early​

Apply the same thinking used during initial explainability work:

  • Build explanations for individual predictions before you build dashboards.
  • Consider which stakeholders need explanations and in what form.
  • Capture explainability outputs alongside the model — treat them as a first-class artefact. This approach helps move work from opaque delivery toward transparent, reviewable systems that stakeholders can actually trust.

Action item: Share how you will apply new skills to your role.​

  • Where in your role could a reproducible, versioned first deployment reduce risk in a project you already run?
  • Which stakeholders in your organisation would benefit most from clearer explanations of automated or model-driven decisions?

Don't know where to start?​

Think about areas in your organisation where:

  • A manual, one-off process could be turned into a versioned, redeployable pipeline.
  • Automated decisions are made but no one can easily explain why a specific outcome happened.
  • A rollback plan would materially reduce the cost of a mistake going live.
In progress