BuildAI.
عربيEN
Start Building
FLAGSHIP PATH · INTERMEDIATE

AI Engineer

Learn to build real AI systems, not just demos.

This journey sequences knowledge step by step, and every stage ends in a capability you can perform.Browse the Knowledge Map →

✓ Production-grade LLM integration✓ Build and evaluate RAG over real documents✓ Tools, agents, evaluation, and cost
Start the journey →Know the basics? Take the placement test →
THE JOURNEY · 8 STAGES
1
STAGE 1AI FoundationsAfter this stage you can explain how an LLM works and judge where AI fits in a product — and where it doesn't.
+
3 concepts · 1 build · 1 challengeStart this stage
  1. 1What is an LLM?Understand text in → text out8 min○
  2. 2TokensControl cost and length4 min○
  3. 3HallucinationDetect and reduce hallucination7 min○
  4. 4Build — A first API call returning a reply, with the key in an env var — not in code.○
  5. 5Challenge — Spot the HallucinationThe bot gives confident answers with invented sources.○
2
STAGE 2LLM ApplicationsAfter this stage you can ship a real streaming, error-handled, cost-budgeted LLM feature.
+
3 concepts · 1 build · 1 challengeStart this stage
  1. 1Conversation stateA chat that remembers6 min○
  2. 2StreamingA snappy experience5 min○
  3. 3Cost controlPredictable cost7 min○
  4. 4Build — AI Chat ApplicationA multi-turn chat with a persona, error handling, and a cost meter.8–12h○
  5. 5Challenge — Context OverflowA chat that answers well for 10 turns then forgets.○
3
STAGE 3Give AI KnowledgeAfter this stage you can build a RAG Assistant that answers from real documents, and diagnose retrieval and hallucination problems.
+
7 concepts · 1 build · 1 challengeStart this stage
  1. 1What is RAG?Understand retrieve → answer8 min○
  2. 2EmbeddingsSearch by meaning6 min○
  3. 3ChunkingSharper retrieval8 min○
  4. 4Hallucination controlTrustworthy answers10 min○
  5. 5Agentic RAGAnswers to compound questions13 min○
  6. 6Advanced RetrieversHigher-precision retrieval13 min○
  7. 7Choosing a RAG stackChoose with confidence10 min○
  8. 8Build — RAG Knowledge AssistantA full RAG system: chunking, semantic search, reranked retrieval, cited answers.15–20h○
  9. 9Challenge — RAG Hallucination HuntConfident wrong answers though the source exists.○
4
STAGE 4Give AI ToolsAfter this stage you can build a system that uses tools safely, with structured outputs and human escalation.
+
5 concepts · 1 build · 1 challengeStart this stage
  1. 1Tool callingA model that acts8 min○
  2. 2Tool securitySafe tools9 min○
  3. 3What is MCP?Understand Server/Client/Host8 min○
  4. 4MCP Clients & HostsBuild a Client / Host12 min○
  5. 5MCP SecuritySafe MCP14 min○
  6. 6Build — Customer Support AgentAn agent using tools with structured outputs and guardrails, escalating with a trace.15–20h○
  7. 7Challenge — Tool Failure CascadeOne tool outage brings the whole agent down.○
5
STAGE 5AgentsAfter this stage you can build an agent that plans, uses tools, and completes multi-step tasks reliably.
+
6 concepts · 1 build · 1 challengeStart this stage
  1. 1The agent loopAn agent that finishes a task10 min○
  2. 2Bounds & budgetAn agent that stops right8 min○
  3. 3The ReAct patternAn agent that reasons out loud8 min○
  4. 4Reflection agentsHigher quality at measured cost9 min○
  5. 5Evaluator–OptimizerOutput that passes the bar9 min○
  6. 6Framework or by hand?Pick the right tool10 min○
  7. 7Build — Research AgentA full agentic loop: planning, search tools, reading, synthesis, citation.18–25h○
  8. 8Challenge — The Agent That Never StopsThe agent loops forever and burns the budget.○
6
STAGE 6Make AI ReliableAfter this stage you can evaluate an AI system honestly and fix its security, cost, and reliability problems.
+
2 concepts · 1 build · 1 challengeStart this stage
  1. 1Evaluating AI systemsNumbers you can defend12 min○
  2. 2Cost engineeringCost under control9 min○
  3. 3Build — An eval set + guardrails + cost engineering on an existing agent.○
  4. 4Challenge — The $40 Agent RunAn agent run costs $40 — cut it to ≤ $4.○
7
STAGE 7Production AIAfter this stage you can reason about running a real AI system: monitoring, cost, and failure behavior.
+
1 concepts · 1 build
  1. 1ObservabilitySee inside the system10 min○
  2. 2Build — A production checkpoint: logging, alerts, spend caps, and a failure plan.○
8
STAGE 8CapstoneAfter the Capstone you have a real, published AI system with evidence and a score — proof you're an AI engineer.
+
1 build
  1. 1Build — A project of your choice: problem → design → build → test → evaluate → improve → publish.○
SKILL MAPAI Engineer
Foundations
AI/ML/GenAITraining vs InferenceLLMsTokensProbabilistic outputHallucinationCapability mapAI vs traditional SWWhere AI fitsResponsible AI
LLM Applications
How LLMs workContext windowChoosing a modelPromptingStructured outputStreaming
Tools
Tool callingThe tool loopTool security
RAG
What is RAGEmbeddingsSemantic searchChunkingAgentic RAGAdvanced RetrieversRAG stack choice
Agents
What is an AgentAgent loopReActReflectionEvaluator–OptimizerFramework choice
MCP
What is MCPMCP SecurityClients & Hosts
Production AI
SecurityReliabilityCost
green skills have pages; the rest soon
Now you can build.
✓Build LLM applications✓Build RAG systems✓Integrate tools✓Build agents✓Evaluate AI systems✓Reason about production
Build your proof — start the Capstone →See the certificates you earn →
AI Engineer · BuildAI