Nymrio Academy
Understand the systems behind intelligent software.
Learn AI from the beginning, then build agents and knowledge systems. Courses with explanations, diagrams, examples and exercises in Romanian and English.
12
77
3
Choose your starting point
Follow the linked courses in order, or return to any topic when you need it.
I am new to AI
I want answers from my documents
I want an AI team
Curriculum
Learn in sequence. Return by topic.
Each track builds one mental model at a time. Follow the complete path or find the lesson that answers the problem in front of you.
12 tracks · 77 lessonsPage 1 of 3
Track 01
AI from the beginning
Understand models, learning and the limits of AI.
F01What is artificial intelligence?AI is a toolbox of methods that help computers solve tasks.F02How does a machine learn?A model learns relationships from examples, like an apprentice seeing many cases.F03What is an LLM and how does it answer?An LLM builds answers token by token.F04Context and memory: what does AI see?Context is the model’s workbench.F05Why can AI be confidently wrong?A well-written answer can hide a wrong fact.F06Chatbot, assistant, agent and agentic systemCommercial names can be unclear.F07Multimodal AI in plain languageMultimodal AI can combine text, images, sound or video.F08Your first useful AI conversationA useful conversation starts with a concrete outcome, not a secret formula.
Track 02
Models, prompts and context
Choose models and build instructions you can verify.
P01Prompt engineering without magic formulasA prompt is a small brief for AI.P02Context engineering: what reaches the model?Context engineering selects what the model sees.P03Temperature, limits and reasoning effortModel settings have tradeoffs.P04How to choose a suitable AI modelThe best model for you succeeds on your tasks at acceptable cost and latency.P05Structured output and tool callingStructured data helps software read an answer, but both structure and meaning require validation..P06Reasoning and verificationA visible plan helps follow a task but is not complete proof of the model’s internal process.P07Prompt, RAG or fine-tuning?These methods change different things.
Track 03
Building and using agents
Turn a conversation into an agent with tools and clear boundaries.
A01Anatomy of an AI agentAn AI agent is more than a model: it is a loop with a goal, tools, state and stopping rules..A02Your first agent in the applicationBuild the Luna Workshop guide from a clear role and a few tests.A03An agent connected to knowledgeAn agent can consult RAG as a knowledge tool.A04An agent using MCP toolsMCP provides capabilities with explicit contracts.A05Web search and code executionWeb search finds public information; code execution can check a calculation.A06Bounded autonomy and human interventionA useful agent does not need unlimited rights.
Track 04
Agentic systems and workflows
Explore all seven orchestration templates.
W01What is an agentic AI system?An agentic system organizes decisions, tools and state toward a goal.W02ReAct: decide, act, observeReAct alternates decisions and actions based on observed results.W03Reflection: improving the first draftReflection resembles a writer and editor.W04Sequential: a team working in orderSequential is a relay: one stage’s output becomes the next stage’s input.W05Orchestrator: coordinator and specialistsOrchestrator chooses suitable specialists and can continue after their results.W06Router: send requests to the right specialistRouter classifies intent and chooses a specialist branch.W07Parallel: several perspectives at onceParallel distributes independent subtasks and then aggregates results.W08Plan-and-Execute: plan, act, adjustPlan-and-Execute separates planning, step execution and plan revision after results.W09Multi-agent: choosing and combining patternsMulti-agent describes collaboration, not a universal switch.