Skip to main content
ANVISoftware Solutions

AI & Generative AI for Developers

Beginner24-30 hours5 modules22 lessons

From what a language model is through to shipping a retrieval-based assistant: prompting, structured output, tools, RAG, agents, evaluation and cost.

Start the first lesson

Prerequisites

  • Ability to program in at least one language
  • Basic understanding of HTTP APIs

Technologies

About this course

This course takes AI from vocabulary to working software. It starts with plain explanations of machine learning, language models, tokens and embeddings, because a great deal of confusion about AI behaviour dissolves once those are clear.

It then moves to building. You will call a model from code, get reliably structured output, let a model use tools, and connect a model to your own documents through retrieval. The emphasis throughout is on treating model output as untrusted input that needs validating.

Evaluation, guardrails, prompt injection and cost control are treated as core engineering topics rather than optional extras, because a feature that is impressive in a demo and unreliable in production has not been finished.

Course roadmap

All 22 lessons are written and available.

Module 01

AI Foundations

The vocabulary, in plain language, with accurate expectations attached.

6/6 ready
  1. 01What Is AI, Really?Distinguish AI, machine learning, deep learning and generative AI, and set accurate expectations.13 min
  2. 02Machine LearningExplain learning from data versus writing explicit rules.14 min
  3. 03Generative AIDescribe what generative models produce and how that differs from classification.12 min
  4. 04Large Language ModelsExplain in plain terms what an LLM does when it responds.15 min
  5. 05Tokens and Context WindowsExplain how tokens and context limits drive cost and behaviour.15 min
  6. 06EmbeddingsDescribe how text becomes vectors and why that enables meaning-based search.14 min

Module 02

Prompting and Structured Output

Get results you can rely on in code, not just results that read well.

4/4 ready
  1. 01Prompt EngineeringWrite prompts that give consistent, usable results.17 min
  2. 02Structured OutputGet schema-conforming output you can validate instead of parsing prose.16 min
  3. 03Tool and Function CallingLet a model invoke your functions, validating everything it returns.19 min
  4. 04Your First Model IntegrationCall a model API from code and handle errors, latency and cost.18 min

Module 03

Retrieval and RAG

Answer questions about your own data without retraining anything.

4/4 ready
  1. 01Why RetrievalExplain why supplying context beats expecting a model to know your data.14 min
  2. 02Vector DatabasesStore embeddings and query them by similarity.16 min
  3. 03Building a RAG PipelineAssemble retrieval and generation into a working question-answering flow.20 min
  4. 04Chunking and Retrieval QualitySplit documents so retrieved context is actually relevant.17 min

Module 04

Agents and Workflows

Multi-step behaviour, and an honest look at when a simple chain is better.

3/3 ready
  1. 01What an Agent IsDescribe multi-step autonomous behaviour and its risks.18 min
  2. 02Agent WorkflowsDesign flows with branching and loops, and know when a chain suffices.19 min
  3. 03Orchestration with LangChain and LangGraphUse orchestration libraries without losing visibility into what runs.18 min

Module 05

Evaluation, Safety and Cost

What turns a convincing demo into something you can actually operate.

5/5 ready
  1. 01Evaluating Output QualityMeasure whether an AI feature is good enough to ship, and keep measuring.19 min
  2. 02GuardrailsConstrain behaviour and handle unacceptable output safely.18 min
  3. 03AI Security and Prompt InjectionDefend against untrusted content reaching a model with your privileges.20 min
  4. 04Cost and LatencyControl token spend and keep responses fast enough to use.17 min
  5. 05Using AI Tools as a DeveloperUse AI assistants to learn faster while verifying everything you accept.16 min

Projects for this course

Advanced

AI Knowledge Assistant

The flagship AI project: a full assistant with retrieval, tool use, conversation, guardrails, evaluation and cost tracking.

35-50 hours
Intermediate

Document Question Answering

Your first retrieval project: ask questions about a set of documents and get answers grounded in them, with citations.

15-20 hours

On completion

What you should take away

  • A language model predicts text; it does not look up facts, which explains most of its failure modes
  • Structured output plus validation is what makes model responses safe to use in code
  • Retrieval is how a model answers questions about your data without retraining
  • Agents add capability and unpredictability together, so scope them deliberately
  • Evaluation, guardrails and cost tracking are part of shipping, not follow-up work

Recommended next course

Software Architecture

Structure systems deliberately: layering, SOLID, clean architecture, patterns, and honest guidance on when each is the wrong choice.

Continue to Architecture

Have a project worth talking through?

Tell us what you're building or what's slowing your current system down. We'll give you a direct read on scope and approach.

Start a ConversationExplore Our Services