Generative AI

Program

Generative AI

Build products powered by large language models

Learn how generative models work, prompt design, retrieval, and how to build applications on top of LLMs.

8

Weeks

40

Chapters

4

Practice formats

1

Portfolio artifact

8 weeks — one module per week (8 modules, 40 chapters), each week ending with a live class every Saturday, plus a module exam before the next module opens.

Curriculum

40 chapters across 8 modules — each with guided notes, a live example, a quiz, an exercise, a challenge, and class practice.

Module 1 · GENERATIVE AI FOUNDATIONS

  1. 1Introduction to Generative AIWhat generative AI is, the kinds of content it creates, how it differs from traditional AI, real uses, and its limitations.Learn →
  2. 2How Large Language Models WorkTokens, embeddings, next-token prediction, transformers and attention, pre-training, fine-tuning, RLHF, and context windows.Learn →
  3. 3Using LLM APIsCalling a model through an API: messages and roles, key parameters, temperature, tokens and cost, streaming, and choosing a model.Learn →
  4. 4Prompt EngineeringWriting clear prompts: instructions, context, examples, output formats, delimiters, step-by-step reasoning, and iterating with tests.Learn →
  5. 5Token Costs, Context Budgets & Model ChoiceHow LLM pricing works, estimating cost per request and per month, input vs. output tokens, choosing model sizes, and routing traffic between models.Learn →

Module 2 · EMBEDDINGS, SEARCH & RAG

  1. 6Embeddings & Semantic SimilarityWhat embeddings are, measuring similarity with cosine, semantic vs keyword search, embedding models, and uses beyond search.Learn →
  2. 7Vector Databases & Hybrid SearchStoring and searching embeddings at scale: vector databases, approximate nearest neighbor indexes, metadata filters, BM25, and hybrid ranking.Learn →
  3. 8Retrieval-Augmented Generation (RAG)Grounding answers in your own data: the RAG pipeline, loading and chunking documents, retrieving context, building the prompt, and citing sources.Learn →
  4. 9Advanced RAG & Retrieval EvaluationQuery rewriting, reranking, and other advanced techniques; measuring retrieval with precision, recall, and MRR; and evaluating answer faithfulness.Learn →
  5. 10Document Parsing & Chunking StrategiesExtracting clean text from documents, fixed-size vs. structure-aware chunking, overlap, chunk size trade-offs, metadata, and testing chunking choices with retrieval metrics.Learn →

Module 3 · TOOLS, AGENTS & CUSTOMIZATION

  1. 11Structured Output & Tool CallingGetting reliable JSON, validating it against a schema, and letting models call functions and APIs through tool use.Learn →
  2. 12AI AgentsThe agent loop, tools, memory, planning, multi-agent patterns, stopping conditions, and keeping agents safe and reliable.Learn →
  3. 13Multimodal Models & Image GenerationVision-language models, diffusion image generation, prompting for images, audio and video models, and responsible use.Learn →
  4. 14Fine-Tuning & Customizing ModelsPrompting vs RAG vs fine-tuning, preparing training data, parameter-efficient fine-tuning (LoRA), evaluation, and costs.Learn →
  5. 15Workflows: Prompt Chaining, Routing & GatesDeterministic LLM workflows vs. autonomous agents, chaining steps, routing by intent, parallel calls, quality gates, and when each pattern fits.Learn →

Module 4 · EVALUATE, SECURE & SHIP

  1. 16Evaluating & Testing LLM ApplicationsWhy evals matter, building test sets, code-based checks, LLM-as-judge, human review, regression testing, and monitoring in production.Learn →
  2. 17Safety, Security & Cost in ProductionPrompt injection, data leakage, guardrails, privacy, bias and responsible AI, and keeping latency and cost under control.Learn →
  3. 18Guardrails & Output ValidationValidating structured outputs, schema checks and retries, business-rule validation, input and output guardrails, and safe fallbacks to humans.Learn →
  4. 19UX Design for AI ProductsSetting expectations, showing sources and uncertainty, editable outputs, feedback controls, latency and streaming, error states, and designing for trust.Learn →
  5. 20Deploying LLM Applications: ArchitectureReference architecture for LLM apps: API gateway, orchestration service, model providers, vector store, caching, queues, secrets, observability, and multi-provider resilience.Learn →

Module 5 · ON THE JOB: BUILDING AI PRODUCT FEATURES

  1. 21On the Job: Scoping an AI FeatureThe CEO wants "AI in the product". Find a problem worth solving, choose between prompting, RAG, fine-tuning or a classic model, estimate cost and risk, and define success before building.Learn →
  2. 22On the Job: A Customer Support Assistant with RAGBuild the tenant help assistant properly — ingest and chunk the knowledge base, retrieve with permissions, answer with citations, say "I don't know", escalate to humans, and measure deflection.Learn →
  3. 23On the Job: Extracting Data from Documents at ScaleProcess 50,000 lease PDFs into structured data — schema design, structured outputs, validation and confidence, human review queues, batch processing, and accuracy measurement.Learn →
  4. 24On the Job: Classification with LLMsUsing LLMs for zero-shot and few-shot classification, label definitions, confusion matrices, per-class precision and recall, and when to switch to a trained classifier.Learn →
  5. 25On the Job: Designing Conversational AssistantsConversation design for chatbots and voice assistants: scope, persona, clarifying questions, memory, handoff to humans, accessibility, and conversation analytics.Learn →

Module 6 · ON THE JOB: QUALITY, SAFETY & TRUST

  1. 26On the Job: Building Evaluation Sets & LLM-as-Judge"Is the new prompt better?" Build representative eval datasets, choose metrics, use code-based checks and LLM graders responsibly, calibrate against humans, and run evals in CI.Learn →
  2. 27On the Job: Red Teaming & Defending Against Prompt InjectionAttack your own AI feature before others do — jailbreaks, direct and indirect prompt injection, data exfiltration via tools, and layered defences that don't rely on the prompt alone.Learn →
  3. 28On the Job: Responsible AI, Privacy & RegulationLegal asks whether the AI features comply. Classify risk, handle personal data properly, document models and limitations, address bias and transparency, and prepare for regulations such as the EU AI Act.Learn →
  4. 29On the Job: Automated Grounding & Hallucination ChecksChecking answers against sources, claim extraction, lexical and number checks, entailment and LLM-as-judge approaches, and combining automated checks with human review.Learn →
  5. 30On the Job: Testing Generative AI for BiasCounterfactual prompt testing, measuring differences across groups, representation in generated content, the four-fifths heuristic, and mitigation steps.Learn →

Module 7 · ON THE JOB: RUNNING LLM APPS IN PRODUCTION

  1. 31On the Job: Cutting LLM Cost & LatencyThe assistant is popular — and costing $13,500 a month with 9-second answers. Measure token usage, trim prompts, cache, route to smaller models, stream responses, and set budgets.Learn →
  2. 32On the Job: Observability & Feedback for LLM AppsA user says the assistant "gave a crazy answer yesterday". Trace every step of LLM requests, capture user feedback, monitor quality signals, and turn production data into improvements.Learn →
  3. 33On the Job: Upgrading Models Without Breaking ThingsA provider deprecates the model you use in 90 days, and a cheaper new model looks promising. Plan migrations, pin versions, run regression evals, shadow test, roll out gradually, and avoid lock-in.Learn →
  4. 34On the Job: Caching & Rate Limiting for LLM AppsExact and semantic caching, cache keys and invalidation, provider rate limits, token-bucket limiting per user, and protecting budgets.Learn →
  5. 35On the Job: Forecasting Generative AI CostsForecasting usage growth, requests per user, cost per request, the effect of caching and routing, budgets, and presenting forecasts to finance.Learn →

Module 8 · CAPSTONE & CAREER LAUNCH

  1. 36Capstone: A RAG ApplicationBuild, evaluate, and present a working RAG application that answers questions grounded in your own documents.Learn →
  2. 37Portfolio & Interview Preparation for AI EngineersStand out as an AI/LLM engineer — build evaluated, deployed projects, explain trade-offs, and prepare for LLM system design, coding and responsible-AI interview questions.Learn →
  3. 38Job Search Strategy for AI EngineersAI engineering role types, what employers look for, shipping portfolio projects with evaluation, writing about your work, and standing out in a fast-moving field.Learn →
  4. 39AI System Design InterviewsHow AI/LLM system design interviews run, a step-by-step framework, discussing retrieval, evaluation, safety, cost and latency trade-offs, and common prompts.Learn →
  5. 40Your First 90 Days as an AI EngineerStart effectively — learn the product, data and existing AI features, set up evals before changing anything, find a measurable improvement, and build trust with product, legal and support teams.Learn →
Start learning →

Delivery

Blended — self-paced lessons plus live build sessions

Portfolio outcome

A working RAG-powered application answering questions grounded in your own documents.

Certificate

Certificate of completion, verifiable online with its credential code.

By the end of this program, you'll be job-ready to…

  • Ground an LLM's answers in real data instead of its own guesses
  • Design prompts and evaluation tests that catch failure modes
  • Build a simple agent that can call tools and take multi-step action

Salary expectations

Estimated U.S. national average annual salary by experience level.

$95,000

Entry-level

$135,000

Mid-level

$180,000

Senior-level

Estimated figures for planning purposes — actual pay varies by location, employer, and experience.

Tools you'll master

Prompt engineeringEmbeddings & vector searchRAG pipelinesAI agents & tool use

Practical training, every chapter

  • A live "try it yourself" example in every chapter you can run immediately
  • A hands-on exercise per chapter, with hints and a reveal-able answer
  • An open-ended challenge per chapter, graded against real success criteria
  • Class practice prompts for pairs or cohorts, not just solo homework

Why take this program

Interview- and job-ready

Built to the same bar Aicloud actually hires against — not a generic syllabus.

Tier 1 company partners

Program partnerships with leading employers are in progress — see Affiliation for status.

Job connection access

Graduates get access to Aicloud's own hiring pipeline and employer referral network.

Professional networking

Cohort discussion boards and live sessions connect you with peers and mentors, not just content.

A real portfolio

You leave with a concrete, explainable artifact — proof of knowledge, not just a certificate.

Register Explore Packages