Agentic AI Learning Path
Sequential roadmap from fundamentals to production
2026 Edition • ~200-400 hours total • Updated February 2026
Progress0 completed / 14 total
How to Use This Roadmap
- →Click steps to mark as completed and track your progress
- →Follow the order — each step builds on previous knowledge
- ★Checkpoints are milestone projects to validate learning
- ⏱Estimated times assume 10-15 hrs/week of focused study
- ⏭Skip steps you already know — they count toward progress but are tracked separately
- ↓Click "Show details" to see learning objectives and resources
Phase 1: Foundations
4-6 weeksPrerequisites: Basic programming experience helpful but not required • Familiarity with command line
0 completed / 3
1
Python & API Basics
BeginnerCore programming skills for AI development
Python 3.10+requests/httpxasync/awaitJSON handlingEnvironment management
2
Prompt Engineering
BeginnerMaster the art of communicating with LLMs
Chain-of-ThoughtFew-shot promptingRole promptingOutput formattingPrompt iteration
3
LLM APIs & Function Calling
★ CHECKPOINTIntermediateConnect to and orchestrate LLM providers
OpenAI APIAnthropic APIFunction/tool callingStreamingError handling
Phase 2: Core Agent Skills
6-8 weeksPrerequisites: Completed Phase 1 • Comfortable with async Python • Basic understanding of LLM APIs
0 completed / 3
4
Tool Use & Integration
IntermediateExtend agents with external capabilities
Tool schemasAPI integrationFile I/OCode executionError recovery
5
Agent Architectures
IntermediateUnderstand core patterns for autonomous agents
ReAct patternPlan-and-ExecuteSelf-reflection loopsDecision policiesGoal decomposition
6
Memory Systems
★ CHECKPOINTIntermediateGive agents persistent context and recall
Conversation memoryVector embeddingsSemantic searchMemory typesContext window management
Phase 3: Production Patterns
6-8 weeksPrerequisites: Completed Phase 2 • Experience building basic agents • Understanding of vector embeddings
0 completed / 3
7
RAG Systems
IntermediateGround agents in your own data
Document chunkingEmbedding modelsVector storesHybrid searchRe-rankingQuery refinement
8
Agent Frameworks
IntermediateLeverage production-ready tooling
LangChainLangGraphCrewAIState machinesGraph-based workflows
9
Orchestration & Workflows
★ CHECKPOINTAdvancedBuild reliable, observable pipelines
DAG designError handlingRetriesConditional logicState managementEvent triggers
Phase 4: Hardening
4-6 weeksPrerequisites: Completed Phase 3 • Working agent in development • Understanding of production systems
0 completed / 3
10
Security & Governance
AdvancedProtect agents and users from harm
Prompt injection defenseInput validationOutput filteringRBACSandboxingAudit logging
11
Monitoring & Evaluation
AdvancedObserve, measure, and improve agent performance
LangSmith/LangfuseTracingMetricsA/B testingHuman feedback loopsBenchmarking
12
Deployment & Scaling
★ CHECKPOINTAdvancedShip agents to production
DockerFastAPIKubernetes basicsRate limitingCachingCost optimization
Phase 5: Advanced
OngoingPrerequisites: Completed Phases 1-4 • Production agent experience • Solid foundation in all core skills
0 completed / 2
★
Multi-Modal & Computer Use
ADVANCEDAdvancedCutting-edge agent capabilities
Vision modelsDocument understandingBrowser automationComputer useAudio/video processing
★
Multi-Agent Systems
ADVANCEDAdvancedCoordinate teams of specialized agents
Agent communicationTask delegationConsensusMCP protocolA2A protocol
Building an AI agent but stuck on deployment?
I can help →