RDL Expert Evaluation Foundations

Turn professional judgment into prompts, quality targets and rubrics that measure where frontier AI models fall short.

What you'll master.

RDL Expert Evaluation Foundations is Rise Data Labs' onboarding curriculum for rubrics projects. Across seven modules you will learn what a rubrics project asks you to produce, how to write prompts that expose meaningful model failures, how to define the Golden Response that sets the quality target, and how to turn that target into criteria another qualified reviewer can apply consistently. The three parts — prompt, quality target and rubric — form the RDL Quality Framework, and every module returns to keeping them aligned. Each module ends with a comprehension check; pass at 70 percent or above to complete it. Fictional-case notice: all organizations, people, data and scenarios used in course examples are fictional and created for instructional purposes.

Module by module.

01

Welcome to Expert Evaluation at RDL

What RDL expects from contributors, what a rubrics project asks you to produce, and the three parts of the RDL Quality Framework — prompt, quality target and rubric — that must stay aligned.

test 20 min
02

What Is an RDL Prompt?

How an evaluation task differs from an ordinary AI request, what counts as a meaningful model failure, and the draft-run-assess loop used to decide whether to keep, revise or replace a prompt.

test 20 min
03

Requirements for Strong Prompts

Eight requirements a prompt must meet, the difference between professional and artificial difficulty, and the nine failure modes that get a prompt rejected.

test 30 min
04

Building Great Prompts

Five construction practices: start from real professional work, build a reasoning path, ground the task in evidence, define the decision and deliverable, and handle assumptions and precision.

test 25 min
05

The Golden Response as a Quality Target

Defining success before scoring it, using the quality target to test the prompt, and diagnosing extraction, reasoning and deliverable failures.

test 20 min
06

What Is a Rubric?

The six rubric components, essential outcomes versus supporting details, evidence chains and dependencies, and four ways a rubric fails even when the analysis behind it is correct.

test 35 min
07

Writing Strong Criterion Descriptions

Eight checks that make a criterion gradable, unambiguous, self-contained, atomic, aligned, stable, precise and fair — plus the final alignment check across the whole task.

test 30 min

Finish certified.

Complete every module and pass each assessment at 70% or above to earn the RDL Expert Evaluation Foundations certificate — downloadable, shareable, verifiable.

apply now