Go from calling an API to shipping a small, reliable AI feature. Learn messages and tokens, structured output, tool calling, retrieval-augmented generation, evaluation, cost control and the security risks unique to LLM apps.
LevelIntermediate
Estimated time28 hours
Syllabus7 modules + capstone
FormatSelf-paced or live cohort
Leads toApprentice rank in AI Engineering
Is this course for you?
Who it’s for
Developers adding AI features to products
Technical analysts and automation builders
Computer science and IT students
Founders prototyping an AI product
Before you start
Comfortable writing basic Python or JavaScript
Know what an API and JSON are
An API key from any LLM provider (usage costs are yours)
What you’ll be able to do
01Call an LLM API with system and user messages, and control temperature and token limits
02Return structured JSON that your code can trust and validate
03Let a model call your own functions (tool calling) safely
04Build retrieval-augmented generation (RAG) with embeddings and a vector store
05Measure quality with a test set and simple evaluations
06Defend against prompt injection and data leaks, and keep costs predictable
Skills for your resume
LLM APIs
Retrieval-augmented generation (RAG)
Embeddings and vector search
Tool and function calling
LLM evaluation
AI application security
Once you’ve earned the rank, list these in the Skills section of your resume and LinkedIn profile.
The syllabus
7 modules · 28 hours · 1,140 XP to collect. Every module ends with a practical project.
Module 1
Your first API calls
+120 XP
Lessons
Messages, roles and tokens
Temperature, max tokens and stop sequences
Streaming responses
Handling errors, timeouts and rate limits
Practical projectBuild a command-line assistant with retries and streaming.
Module 2
Structured output
+120 XP
Lessons
Asking for JSON reliably
Schemas and validation
Repairing or retrying bad output
When to use plain text instead
Practical projectTurn messy support emails into validated JSON tickets.
Module 3
Tool calling
+120 XP
Lessons
How function and tool calling works
Designing safe tool definitions
Multi-step tool use
Keeping a human approval step
Practical projectGive a model a calendar lookup tool and a booking tool that needs confirmation.
Module 4
Embeddings and retrieval (RAG)
+120 XP
Lessons
What embeddings are
Chunking documents well
Vector search and ranking
Grounding answers and citing sources
Practical projectBuild a question-answering bot over a small document set, with citations.
Module 5
Evaluation and quality
+120 XP
Lessons
Building a test set
Automatic checks and model-graded checks
Regression testing when prompts change
Logging and tracing
Practical projectWrite 30 test cases and score two prompt versions against them.
Module 6
Security for LLM apps
+120 XP
Lessons
Prompt injection, direct and indirect
Sensitive data exposure
Excessive agency and unsafe tool use
The OWASP Top 10 for LLM Applications
Practical projectRed-team your own RAG bot and fix three weaknesses you find.
Module 7
Cost, speed and shipping
+120 XP
Lessons
Counting tokens and estimating cost
Caching and choosing smaller models
Latency and user experience
Deploying a small AI feature
Practical projectAdd cost tracking and a cache to your bot and report the savings.
Capstone project · +300 XP
Show what you can do
Ship a small AI feature end to end: a RAG or tool-using assistant for a realistic use case, with structured output, a 30-case evaluation, documented security controls and a cost estimate. Submit the code repository link and a short write-up.
Live viva, at launch. A recorded 5–10 minute walk-through with an assessor or teacher who can ask follow-up questions.
Process evidence. Submit drafts, notes or commit history, not just a finished file.
AI disclosure. If you used AI, say how. Hidden use can void the project.
Checks are a signal, not the verdict. AI-detection and similarity reviews help the assessor. They are never the only proof.
XP is just for fun. It tracks your progress, but it can never buy or skip a rank.
The Rank Trial
When you’ve finished every module and the capstone, the Apprentice Trial in AI Engineering opens. The Guild runs it, not your instructor, so the rank means the same whoever taught you. It is a proctored, closed-book exam.
What’s assessed
API concepts: tokens, roles and parameters
Structured output and tool calling
RAG design and evaluation
LLM security and cost control
How it works
Format
Proctored online exam, closed notes
Questions
45, randomised bank of scenarios, lab tasks and short answers in your own words
Time
75 minutes, timed
Pass mark
70%
Identity
Verified account, ID check at launch
Re-sit
After 7 days if you don’t pass; a voided attempt may wait longer
AI in lessons
This is an AI course, so you will use AI tools in the lessons and projects where the module says so. That is part of the skill.
AI in the Trial
Never. The Trial is closed-book and proctored. No assistants, notes or other people.
AI in projects
You may use it only with a clear disclosure of what the tool did and what you did yourself.
Integrity · at launch
Built to stop easy AI cheating
Proctoring, identity checks and live labs start at launch. These are the rules the platform will enforce.
Webcam and screen shareYou share your camera and your screen for the whole sitting.
Locked-down browserThe Trial runs in a locked browser. Copy and paste, extra screens and leaving the tab are flagged.
ID checkYou sit it from a verified Guild account, with an identity check before the clock starts.
Randomised question bankEach attempt draws a fresh mix of questions, so a leaked paper or a chatbot answer key will not help.
Time-boxedThe sitting is timed. When the clock ends, the attempt is submitted as it stands.
Live sandbox labsPractical tasks use a live lab with data unique to you, not a question you can paste into an AI tool.
Ranks are earned. They can’t be bought, and no amount of XP can skip a Trial. Cheating voids the attempt and can revoke a rank already awarded.
Your path in AI Engineering
ApprenticeThis course
JourneymanCourses coming
ExpertCourses coming
MasterCourses coming
The credential you earn
Example
Guild credential
Apprentice · AI Engineering
Proctored
Awarded to
Your name
Course
Building with LLM APIs
Credential ID
GS-XXXX-XXXX
Awarded by
Guild of Skills
Assessment
Proctored Rank Trial
Check it at
guildofskills.com/verify
Example only. Each real rank gets its own ID, a Proctored badge and a verify page.
On your resume
Add it under Licences & Certifications: Apprentice, AI Engineering · Guild of Skills, the month you passed, and your credential ID so employers can check it.
On LinkedIn
Open your profile and add a Licence or certification
Name: Apprentice, AI Engineering. Issuing organisation: Guild of Skills
Add the issue date, your credential ID and your verify link
Add to LinkedInUnlocks when you pass the Trial
Your instructor
Guild-approved instructor coming soon
Every instructor is reviewed and approved by the Guild Council before they can teach, and every lesson and video link is checked before students see it.
Provider-neutral: the patterns apply to all major LLM APIs. Includes the OWASP Top 10 for LLM Applications. A Guild credential is not a vendor certification.
Questions about this course
Which AI provider do I use?
Any major provider with an API works. Examples are written to be provider-neutral, and you pay the provider directly for your own API usage.
How much does it cost?
Pricing is announced at launch. Some courses may be free and some paid.
When can I start?
Enrolment is open now. Create a free account, confirm your email, then enrol from this page. Lessons, pricing and live cohorts are confirmed at launch.
Do I get a certificate just for finishing?
Finishing the course unlocks the Rank Trial. Your rank, certificate and verify page are awarded only when you pass the Trial.
Is this the same as an industry certification?
No. A Guild rank is a Guild of Skills credential with its own published standard and public verify page. It isn’t a vendor certification and doesn’t include any vendor exam.
Can I use AI in the Rank Trial?
No. The Trial is closed-book and proctored. AI assistants, notes and other people are not allowed. A flagged attempt can be voided.
Can I use AI while I learn?
Yes, where a module says so, especially in the AI courses, where using the tools is part of the skill. You must disclose any AI you use in a project or capstone.
What is a viva?
After you submit your capstone you walk an assessor or teacher through it, live, for five to ten minutes, and answer follow-up questions. The session is recorded. It starts at launch.
What happens if I cheat?
The attempt is voided. You wait to re-sit, at least seven days, and longer for a serious breach. A rank already awarded can be revoked, and the public verify page will show that.
Leads toApprentice rank in AI Engineering
JourneymanCourses coming
ExpertCourses coming
MasterCourses coming
