Agentic AI: Building with LLM Agents

Program demonstrativ 2026 Tehnologie Avansat en 440 min

A hands-on, advanced course on building software with LLM agents. Learn how large language models actually work, master prompting and function calling, design agent loops with planning and reflection, add RAG and memory, orchestrate multi-agent systems, and ship safe, evaluated agents to production — guided by talks from Andrej Karpathy, Anthropic, OpenAI, LangChain and Andrew Ng.

From language model to autonomous agent

Large language models stopped being chatbots the moment we gave them tools. This advanced course takes developers from a solid mental model of how LLMs actually work — tokens, training pipelines, prompting — to the engineering patterns behind production agentic systems: function calling, planning loops, retrieval-augmented generation, memory, and multi-agent orchestration.

Each lesson pairs a carefully selected talk from the people who built this field — Andrej Karpathy, Anthropic and OpenAI engineers, Harrison Chase of LangChain, Jerry Liu of LlamaIndex, Andrew Ng — with structured reading notes, curated references, and quizzes that test real understanding.

What you will be able to do

  • Explain the LLM training pipeline (pretraining, SFT, RLHF) and its practical consequences
  • Design and version production-grade prompts
  • Implement tool use / function calling safely and reliably
  • Build agent loops with ReAct-style reasoning, planning and reflection
  • Architect RAG pipelines and agent memory that survive production traffic
  • Decide when (and when not) to use multi-agent architectures
  • Evaluate, guard and ship agents with proper safety controls

Prerequisites: comfortable programming in at least one language and basic familiarity with calling web APIs. No machine learning background required.

Structura cursului

13 lecții
01 Foundations: How Large Language Models Work 3 lecții · 54 min
  • Inside a Large Language Model 18 min
  • Test LLM Foundations Check
  • The Training Pipeline: From Pretraining to Assistant 18 min
  • Prompt Engineering for Builders 18 min
02 From Chatbot to Agent: Tools, Function Calling and Planning Loops 3 lecții · 54 min
  • What Makes an Agent? Agentic Workflow Patterns 18 min
  • Tool Use and Function Calling 18 min
  • Test Tool Use & Function Calling Check
  • The Agent Loop: ReAct, Planning and Reflection 18 min
03 Knowledge and Scale: RAG, Memory and Multi-Agent Systems 4 lecții · 64 min
  • Retrieval-Augmented Generation: Fundamentals 18 min
  • Test RAG Essentials — Quick Check
  • Production-Grade RAG 18 min
  • Agent Memory: Short-Term and Long-Term 10 min
  • Multi-Agent Architectures 18 min
04 Trustworthy Agents: Evaluation, Safety and Production 3 lecții · 54 min
  • Evaluating and Optimizing LLM Systems 18 min
  • Safety and Alignment for Agentic AI 18 min
  • Shipping Agents: Lessons from the Field 18 min
Evaluare finală
  • Test Final Exam: Agentic AI
Autentificați-vă pentru a vă înscrie