AI Agents
Outline34 topics and 63 subtopics mapped. This is the ground the track will cover — the lessons aren't written yet.
- 01
Basic Backend Development
- 02
Git and Terminal Usage
- 03
REST API Knowledge
- 04
Transformer Models and LLMs
8 subtopics- Tokenization
- Context Windows
- Token Based Pricing
- Temperature
- Frequency Penalty
- Presence Penalty
- Top-p
- Stopping Criteria
- 05
Open Weight Models
- 06
Closed Weight Models
- 07
Streamed vs Unstreamed Responses
- 08
Reasoning vs Standard Models
- 09
Fine-tuning vs Prompt Engineering
- 10
Embeddings and Vector Search
- 11
Understand the Basics of RAG
- 12
Pricing of Common Models
2 subtopics- What are AI Agents?
- What are Tools?
- 13
Agent Loop
10 subtopics- Perception / User Input
- Reason and Plan
- Acting / Tool Invocation
- Observation & Reflection
- Personal assistant
- Code generation
- Data analysis
- Web Scraping / Crawling
- NPC / Game AI
- Max Length
- 14
What is Prompt Engineering
6 subtopics- 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 etc
- 15
Tool Definition
6 subtopics- Web Search
- Code Execution / REPL
- Database Queries
- API Requests
- Email / Slack / SMS
- File System Access
- 16
What is Agent Memory?
- 17
Short Term Memory
- 18
Long Term Memory
- 19
Episodic vs Semantic Memory
4 subtopics- RAG and Vector Databases
- User Profile Storage
- Summarization / Compression
- Forgetting / Aging Strategies
- 20
ReAct (Reason + Act)
- 21
Model Context Protocol (MCP)
3 subtopics- MCP Hosts
- MCP Client
- MCP Servers
- 22
Creating MCP Servers
2 subtopics- Local Desktop
- Remote / Cloud
- 23
Chain of Thought (CoT)
- 24
RAG Agent
- 25
Planner Executor
- 26
DAG Agents
- 27
Tree-of-Thought
- 28
Manual (from scratch)
- 29
LLM Native "Function Calling"
22 subtopics- OpenAI Functions Calling
- Gemini Function Calling
- OpenAI Assistant API
- Langchain
- LlamaIndex
- Haystack
- AutoGen
- CrewAI
- Smol Depot
- Anthropic Tool Use
- Metrics to Track
- Unit Testing for Individual Tools
- Integration Testing for Flows
- Human in the Loop Evaluation
- LangSmith
- Ragas
- DeepEval
- Structured logging & tracing
- LangSmith
- Helicone
- LangFuse
- openllmetry
- 30
Prompt Injection / Jailbreaks
- 31
Tool sandboxing / Permissioning
- 32
Data Privacy + PII Redaction
- 33
Bias & Toxicity Guardrails
- 34
Safety + Red Team Testing
