Roadmap.sh Alignment
This page maps the roadmap.sh AI Agents roadmap into this repository's learning-database structure.
Prerequisites
Repository stage: 01 Prerequisites
- Basic Backend Development
- Git and Terminal Usage
- REST API Knowledge
- Streamed vs Unstreamed Responses
LLM Fundamentals
Repository stage: 02 LLM Fundamentals
- Transformer Models and LLMs
- Tokenization
- Context Windows
- Token Based Pricing
- Open Weight Models
- Closed Weight Models
- Reasoning vs Standard Models
- Fine-tuning vs Prompt Engineering
- Embeddings and Vector Search
- Understand the Basics of RAG
- Pricing of Common Models
Generation Controls
Repository stage: 02 LLM Fundamentals
- Temperature
- Top-p
- Frequency Penalty
- Presence Penalty
- Stopping Criteria
- Max Length
Prompt Engineering
Repository stage: 03 Prompt Engineering
- What is Prompt Engineering
- Be specific in what you want
- Provide additional context
- Use relevant technical terms
- Use examples in your prompt
- Iterate and test your prompts
- Specify length, format, and output constraints
AI Agents 101
Repository stage: 04 Agent Fundamentals
- What are AI Agents?
- What are Tools?
- Agent Loop
- Perception / User Input
- Reason and Plan
- Acting / Tool Invocation
- Observation and Reflection
Tools and Actions
Repository stage: 05 Tools and Actions
- Tool Definition
- Name and Description
- Input / Output Schema
- Error Handling
- Usage Examples
- Web Search
- Code Execution / REPL
- Database Queries
- API Requests
- Email / Slack / SMS
- File System Access
MCP
Repository stage: 06 MCP
- Model Context Protocol
- MCP Hosts
- MCP Client
- MCP Servers
- Creating MCP Servers
- Local Desktop
- Remote / Cloud
Agent Memory
Repository stage: 07 RAG and Memory
- What is Agent Memory?
- Short Term Memory
- Long Term Memory
- Within Prompt
- Vector DB / SQL / Custom
- Episodic vs Semantic Memory
- RAG and Vector Databases
- User Profile Storage
- Summarization / Compression
- Forgetting / Aging Strategies
Agent Architectures
Repository stage: 08 Agent Architectures
- ReAct (Reason + Act)
- RAG Agent
- Chain of Thought
- Planner Executor
- DAG Agents
- Tree-of-Thought
Building Agents
Repository stage: 09 Building Agents
- Manual from scratch
- Direct LLM API calls
- Implementing the agent loop
- Parsing model output
- Error and rate-limit handling
- LLM Native Function Calling
- OpenAI Function Calling
- Gemini Function Calling
- Anthropic Tool Use
- LangChain
- LlamaIndex
- Haystack
- AutoGen
- CrewAI
Evaluation and Observability
Repository stage: 11 Evaluation and Observability
- Metrics to Track
- Unit Testing for Individual Tools
- Integration Testing for Flows
- Human in the Loop Evaluation
- LangSmith
- Ragas
- DeepEval
- Structured logging and tracing
- Helicone
- LangFuse
- OpenLLMetry
Security and Ethics
Repository stage: 12 Security and Ethics
- Prompt Injection / Jailbreaks
- Tool Sandboxing / Permissioning
- Data Privacy + PII Redaction
- Bias and Toxicity Guardrails
- Safety + Red Team Testing
Repository Additions Beyond Roadmap.sh
These sections are added to make the site more useful as a professional learning database:
- 10 Multi-Agent Systems
- 13 Production Deployment
- Reference Library
- Topic-level resource collections
- Stage checkpoints and measurable exit criteria