All academies
FoundationsBeginnerFoundation certificate

Prompt Engineering Academy

Make models do exactly what you meant.

Prompting stops being guesswork once you can measure it. You'll build a prompt library, an evaluation harness and a regression suite — the same discipline you'd apply to any other production interface.

Duration

4 weeks

Content

32 hours

Lessons

42

Level

Beginner

Students

1,240

Certified

812

4.9average ratingEnrolling now

RIVA mentors you through every lesson, lab and assessment in this academy — and knows where you got stuck last time.

What you’ll be able to do

Written as capabilities, not topics. If you can’t do these at the end, the academy didn’t work.

  • Design system prompts that hold under adversarial input
  • Build a reusable, versioned prompt library with evaluation baked in
  • Measure prompt quality with graders instead of vibes
  • Cut token spend without losing output quality

Skills you'll build

System promptsFew-shot designStructured outputPrompt evaluationGuardrails

Tools you'll use

ClaudeOpenAIPromptfooJSON Schema
Curriculum

5 modules · 21 lessons

About 11 hours of content. Expand a module to see every lesson.

01How models read a promptTokens, context and attention — enough mechanics to stop guessing.
  • VideoTokens, context windows and costWhy the same request costs 3× depending on how you phrase it.14 min
  • VideoWhat the model actually attends toPosition, recency and the lost-in-the-middle effect.16 min
  • VideoTemperature, top-p and when to pin them to zero12 min
  • LabLab: measuring token spend on a real prompt25 min
02System prompts that holdThe instructions that survive an adversarial user.
  • VideoAnatomy of a production system promptRole, constraints, tools, refusals — in that order, and why.18 min
  • VideoInstruction hierarchy and conflict resolution15 min
  • VideoRefusals, escalation and the honest 'I don't know'13 min
  • LabLab: harden a system prompt against injection30 min
  • QuizCheckpoint quiz8 min
03Structured outputGetting JSON you can parse, every time, without a retry loop.
  • VideoSchema-first prompting16 min
  • VideoTool calling as a typing disciplineWhy a tool definition beats 'respond in JSON' every time.18 min
  • VideoRecovering from malformed output without a retry storm14 min
  • LabLab: typed extraction from messy PDFs35 min
04EvaluationThe part that separates a prompt engineer from someone with opinions.
  • VideoBuilding an eval set from real traffic20 min
  • VideoGraders: exact match, rubric, and LLM-as-judge22 min
  • VideoRegression gates in CIFailing a build on a quality drop, not just on a stack trace.16 min
  • LabLab: wire promptfoo into a pipeline32 min
  • QuizModule quiz10 min
05CapstoneShip a versioned prompt library with evaluation attached.
  • VideoDesigning the library surface15 min
  • LiveLive build session with a mentor60 min
  • ProjectCapstone: prompt library + eval harnessAssessed against the Foundation rubric. Passing issues your certificate.240 min

Hands-on labs

Build a prompt evaluation harness

Score a prompt against a small golden set and fail the run when quality drops below a threshold — the seed of a CI gate.

45 min · Beginner · runs in the browser

Projects you can ship

On completion

ReveloSoft Certified Foundation

You understand the vocabulary and can follow a working system end to end. The rung that makes the rest of the ladder legible.

60-minute assessment · pass mark 70% · QR-verifiable credential

  • Complete one Foundation academy in full
  • Pass the 40-question proctored assessment
  • Submit one guided project
How certification works