You now know what AI, a model, and an LLM are, and the ideas behind almost every AI product: models predict tokens, everything must fit in the context window, settings like temperature shape the output, structured output turns answers into data, embeddings capture meaning, grounding keeps answers honest, and tools turn models into agents.
That is the vocabulary. The next step is building real things with it — and seeing where they break.
Quick recap
| Idea | One-line summary |
|---|---|
| AI, ML, generative AI | AI is the umbrella; ML learns from examples; generative AI creates new content |
| Model | A trained set of numbers that turns an input into an output |
| LLM | A large language model, trained on huge amounts of text |
| Prompt and API | Your app sends a prompt to the provider's API — and keeps the API key secret |
| Open vs closed, cloud vs on-device | Download-and-run versus API-only; big server models versus small phone models |
| Language model | Predicts the next token, one step at a time |
| Token | The small chunk of text models read, count, and charge for |
| Context window | Everything the model can see in one request — including the conversation |
| Temperature | Low for consistent facts, higher for creative variety |
| Structured output | Ask for JSON, then always validate it |
| Embedding | Text turned into numbers so similar meanings are close |
| Grounding / RAG | Give the model real documents and let it say "not found" |
| Tools and agents | The model asks; your code decides and runs |
Choose your path
All three paths are free and self-paced. They share one Python course, so nothing is wasted if you change your mind.
- AI Engineer — build AI applications that hold up in production: retrieval, evaluation, security, and deployment. Start here if unsure.
- Agentic AI — build agents that use tools, remember context, recover from failures, and ask a human before acting.
- Forward Deployed Engineer — take AI into real companies, from discovery and integration to pilot and handoff.
Not sure which fits? Compare the three paths.
Already an Android developer?
Put these ideas straight into an app. The Android + AI codelabs walk you through Gemini chat, structured output, on-device models, and live voice — step by step, in Kotlin.
Want help planning your move into AI?
Book a 1:1 session with Akshay, or start with the Android AI Interview Handbook if you are preparing for interviews.