“AI is still new to me.”
Start here if: you cannot yet clearly distinguish AI, machine learning, neural networks and deep learning, or explain what an intelligent agent observes, decides and does.
Begin Module 11What can I explain without relying on product names?
2Which prerequisite becomes unclear when I follow the reasoning?
3Do I need foundations, system context or build practice next?
Choose by starting point
Which statement sounds most like you?
Choose the first statement whose readiness check you cannot yet satisfy. Starting earlier is useful when a missing concept blocks later reasoning; starting later is reasonable when you can explain the prerequisites and verify them through examples.
“I understand AI ideas, but the technical pieces are weak.”
Start here if: AI concepts make sense, but Python, data structures, algebra, probability, statistics, development tools or reproducible experiments interrupt your progress.
Open Module 2“I know the foundations and want to understand today’s models.”
Start here if: you can explain training versus inference, common learning paradigms, neural-network basics and evaluation splits, but need a system-level map of models and large-language-model ecosystems.
Explore AI Models“I want to understand chatbots, assistants and agents.”
Start here if: you understand model inference, tokens, embeddings and evaluation basics and now need to connect them to conversation, RAG, memory, tools, agents and operations.
Explore Assistants“I want to make sense of the wider AI tool landscape.”
Start here if: your goal is to choose tools responsibly and you can separate a product feature from the underlying model, workflow, data, integration and risk decisions.
Explore AI Tools“I understand enough theory. I need implementation practice.”
Start here if: you can work with Python, packages, environment variables, JSON and HTTP APIs and can explain the AI pattern you intend to implement and evaluate. Guided Labs are the safest default build entry; if you already satisfy a later track’s published readiness criteria, you may select Portfolio Projects or AI Build Challenges directly.
Start a Guided LabCore foundation path
What each of the five Modules unlocks
The Modules are sequential enough to form a curriculum, but modular enough to revisit as reference material.
1AI Foundations
Build the conceptual map and language used across the site.
12 Chapters 2Technical FoundationsBuild the programming, data, mathematics and systems base that later topics assume.
12 Chapters 3Machine LearningUnderstand learning paradigms, workflows, evaluation, features and model development.
12 Chapters 4Neural NetworksUnderstand neurons, network architectures, training and optimisation mechanisms.
12 Chapters 5Deep LearningConnect deeper neural architectures and representation learning to modern AI capabilities.
12 ChaptersSuggested routes
Four ways to enter and navigate NIYAMA-VIDYĀ
The ranges below are planning estimates for focused self-study, not deadlines or completion promises. Reading speed, prerequisite gaps, exercises and build depth can change the actual time substantially.
Foundation-first learner
M1→M2→M3→M4→M5→Explore
Suggested range: 6–12 months at roughly 4–6 focused hours per week. Best when you want the complete foundation sequence and time to complete exercises rather than only read.
Modern AI systems learner
Foundations as needed→Models→Assistants→Tools
Suggested range: 8–14 weeks at roughly 4–6 focused hours per week. Best when you can pass the foundation checks and need a current model, assistant and tool-system map.
Assistant builder
Models→Assistants→Labs→Projects→Challenges
Suggested range: 10–16 weeks at roughly 5–8 focused hours per week. Best when you can explain the model/RAG/tool concepts and want guided implementation, portfolio evidence and independent build proof. Experienced learners may enter a later build track when its published readiness criteria are already satisfied.
Practitioner / decision-maker
Explore selectively→Resources→Build where useful
Suggested range: 4–8 weeks at roughly 3–5 focused hours per week for a selected question. Best for targeted evaluation; return to prerequisites whenever a decision depends on mechanisms you cannot explain.
A useful rule
Skip confidently—but return when a concept becomes a blocker.
Do not relearn what you already know. Use headings, summaries and resource pages to test whether a prerequisite is actually missing.
Do not treat code as theory. If a lab uses a concept you cannot explain—RAG, embeddings, tool calling, evaluation—follow the linked Explore material first.
Do not confuse product familiarity with system understanding. Knowing how to use an app is different from understanding why the architecture behaves as it does.
Completion credential note. Selective routes are valid for self-directed learning, but skipped items do not count as completed. The current NIYAMA-VIDYĀ Full-Library Certificate of Completion requires explicit completion of all 125 registered learning items.
Need the whole map?
The AI Ecosystem page is NIYAMA-VIDYĀ’s master blueprint.
Use it to see the complete learning architecture, how the layers connect and where Resources supports the journey.