Skip to content
NTPU AI Learn

DS 302 Apply

Turn what you learned into your own Data Agent Architecture Canvas + Governed Prototype + Evaluation Pack

What this final artifact is for

The Data Agent Architecture Canvas + Governed Prototype + Evaluation Pack is DS 302’s final artifact. Pick a Data Agent you actually need, design it with governed tool access, a trusted semantic layer, a verification harness, and a human approval point, and write clearly about the problem, audience, data, method, verification, evidence, limitations, risk and governance, human reviewer, and next step. This becomes the starting point for later governance discussions and implementation decisions.

Info

Human, AI, Human: a person decides whether an agent is justified and defines the problem, audience, and quality expectations; the agent plans and executes within governed bounds; a person verifies the evidence, claims, and the decision to publish or use the result, and remains responsible for it.

What to fill in

  • Problem, audience, and data (source systems, grain, time coverage, key definitions, dataset version)
  • Method and workflow (covering intent → plan → tool/data access → execute → verify → explain → approve → observe)
  • At least three verification checks
  • Evidence and output: the verified answer this agent actually produced and its evaluation results
  • Limitations
  • Risk and governance considerations (permissions, threat model, human approval, observability)
  • Human reviewer
  • Next step

A well-structured Data Agent Architecture Canvas + Governed Prototype + Evaluation Pack

A well-structured artifact includes: a clear, testable problem with a stated reason an agent is justified; a clearly defined data source, grain, and dataset version; a reproducible method; at least three verification checks; evidence and claims that match the observed, verified output; and at least one limitation.

Info

The platform does not automatically judge whether your agent architecture is entirely correct. This artifact is used for learning feedback and later discussion, and is reviewed by a human.

Exit Ticket

Finally, use the Exit Ticket to record today’s most important idea, the step you are still unsure about, where you will apply this method in a real context, the risk you most need to verify, and your difficulty and usefulness ratings.

Before you submit

Warning

Do not enter company secrets, client data, personal data, or other unauthorized sensitive information. Unless authorized, never upload confidential organizational data to an AI tool; check your organization’s policy before using real operational data.

Go to the forms

What you need to complete

  • Complete both DS 302 modules
  • Score at least 80% on the course post-test
  • To earn the Data Agents & Analytics Capability Certificate, also submit a Data Agent Architecture Canvas + Governed Prototype + Evaluation Pack and have it reviewed and passed by a Credential Admin against the rubric

How you are assessed

A Credential Admin scores the Data Agent Architecture Canvas + Governed Prototype + Evaluation Pack against eight weighted domains: Problem & Agent Justification; Data Grounding; Method / Workflow Completeness; Verification Rigor; Evidence–Claim Consistency; Limitations Disclosure; Risk & Governance; and Transferable Method & Evaluation Pack. A score of at least 19 out of 24, with no critical-fail rule triggered, is a PASS; 15 or below, or any triggered critical-fail rule, is a FAIL.

Info

This scoring logic is published so you know what is being assessed. It is not the same as revealing quiz answers.

Practical evidence

  • A recorded rubric result (PASS, REVISION_REQUIRED, or FAIL) with no critical-fail rule triggered
  • A PASS records capability evidence separately for each of the nine fine-grained DS 302 competencies (agent-vs-workflow justification, analysis planning, tool contract design, semantic layer grounding, permission minimization, verification before answering, golden-question evaluation & prompt-injection defense, observability & failure recovery, and governed agent delivery)

What you can earn

DS 302 Course Completion Certificate: earned once you complete both modules and meet the post-test threshold, with no Data Agent Architecture Canvas + Governed Prototype + Evaluation Pack submission required. Data Agents & Analytics Capability Certificate: requires a separately submitted and reviewed final artifact. The two are independent and do not affect each other.

What this credential does not mean

This credential confirms the holder completed DS 302's required course and post-test, or (for the certificate) demonstrated tool design, verification, risk governance, and evaluation capability on the Data Agent Architecture Canvas + Governed Prototype + Evaluation Pack.

  • It does not mean the holder holds a formal data engineering or governance professional license or is qualified for a specific role
  • It does not mean any specific Data Agent the holder designs later is automatically correct or safe
  • It does not mean human review can be skipped before a high-impact decision (Human, AI, Human)

What next?