Curriculum

Modules → topics → lessons. Click a topic to see its lessons — advance by finishing each topic.

Foundations: The AI Engineer & LLMs

The AI engineer role & model landscape3 lessonsupcoming

·What an AI engineer is (vs ML engineer) & career pathsexplainer

·The 2026 model landscape: closed vs open modelscase-study

·Platforms & ecosystem: Hugging Face, OpenRouter, Ollama, gatewaysexplainer

How LLMs actually work2/3 lessonsin progress

✓Tokens, tokenization & context windowsexplainer

·Next-token prediction, attention & why LLMs hallucinateexplain-back

✓Sampling: temperature, top-p/top-k, repetition penaltiescode-along

Calling models from code3 lessonsupcoming

·The chat API: messages, roles, system promptscode-along

·Streaming responses & token/cost accountingcode-along

·Multi-provider code: OpenAI-compatible APIs & gateway routingcase-study

Prompt & Context Engineering

Prompt fundamentals3 lessonsupcoming

·Clear instructions, zero-shot & few-shotcode-along

·Chain-of-thought & reasoning modelsexplain-back

·Role/system prompting & prompt templatescase-study

Reliable, structured output3 lessonsupcoming

·Structured outputs & JSON schema enforcementcode-along

·Prompt caching & long-prompt economicsexplainer

·Debugging flaky prompts: prompt regression testingdebug

Context engineering3 lessonsupcoming

·The four moves: write, select, compress, isolateexplainer

·Session memory, compaction & summarizationcode-along

·Knowledge injection & dynamic context compositioncase-study

Embeddings & Vector Search

Embeddings3 lessonsupcoming

·What embeddings are: meaning as vectorsexplainer

·Similarity, cosine distance & choosing an embedding modelcode-along

·Beyond search: classification, clustering, recommendationscase-study

Vector databases & search3 lessonsupcoming

·kNN vs ANN: how HNSW indexing worksexplainer

·Vector DBs hands-on: pgvector, Chroma, Qdrantcode-along

·Metadata filtering & hybrid (keyword + vector) searchcode-along

RAG: Basics to Production

RAG fundamentals3 lessonsupcoming

·The retrieve→augment→generate loop & RAG vs fine-tuningexplainer

·Chunking strategies: size, overlap, semantic & structuralcode-along

·Grounded answers with citationscode-along

Production RAG4 lessonsupcoming

·Query rewriting, HyDE & multi-query retrievalexplainer

·Reranking & hybrid retrievalcode-along

·RAG evaluation: faithfulness, relevance & hallucination mitigationcase-study

·Debugging retrieval failure modesdebug

Tool Use & Agents

Tool calling2 lessonsupcoming

·Function/tool calling: schemas & the tool-use loopexplainer

·Executing tools, parsing results, errors & retriescode-along

Agent loops3 lessonsupcoming

·From workflows to agents: ReAct & plan-and-executeexplainer

·Stopping conditions, reflection & self-correctionexplain-back

·Agent failure modes: loops, tool-thrashing, runaway costdebug

Agent memory2 lessonsupcoming

·Short-term vs long-term memory designexplainer

·Compaction, external memory stores & retrieval-backed memorycode-along

Agent Frameworks, MCP & Multi-Agent

Agent frameworks3 lessonsupcoming

·The 2026 framework landscape: LangGraph, OpenAI Agents SDK, Claude Agent SDKcase-study

·Build the same agent twice: LangGraph vs an SDKcode-along

·When to skip frameworks entirelyexplain-back

Model Context Protocol (MCP)3 lessonsupcoming

·MCP architecture: hosts, clients, servers, transportsexplainer

·Building your own MCP server (tools + resources)code-along

·MCP security: injection via tools, permissions, trust boundariescase-study

Multi-agent systems2 lessonsupcoming

·Orchestrator + sub-agents, handoffs & context isolationexplainer

·When multi-agent helps (and when it just costs more)case-study

Evals, Observability & Safety

Evals3 lessonsupcoming

·Evals are unit tests: golden datasets & assertion-based evalscode-along

·LLM-as-judge: rubrics, pairwise comparison, judge biascase-study

·Evals in CI + turning production failures into eval casescode-along

Observability3 lessonsupcoming

·Tracing LLM apps: spans across prompts, tools & retrievalcode-along

·Cost & latency engineering: caching, routing, batching, tieringexplainer

·Monitoring quality drift in productioncase-study

Safety & guardrails3 lessonsupcoming

·Prompt injection & jailbreaks: attack and defend your own botdebug

·Guardrails: input/output validation, PII handling, moderationcode-along

·Responsible AI: bias, disclosure, human-in-the-loopexplain-back

Shipping: APIs & Deployment

AI backends3 lessonsupcoming

·FastAPI + async Python for LLM appscode-along

·Streaming endpoints (SSE/WebSockets) & durable background jobscode-along

·Auth, rate limiting & API design for AI productsexplainer

Deployment & infra3 lessonsupcoming

·Docker for AI apps & environment managementcode-along

·Deployment targets: serverless vs containers; gateways & failovercase-study

·Rollouts: feature flags, canary prompts, prompts versioned like codeexplainer

AI-assisted engineering2 lessonsupcoming

·Claude Code, Cursor & agentic coding workflowscode-along

·Working with coding agents: specs, skills, review disciplineexplain-back

Open Models, Fine-Tuning & Multimodal

Open models & local inference2 lessonsupcoming

·Running open models: Ollama locally, vLLM for servingcode-along

·Quantization: GGUF, 4-bit, quality/VRAM trade-offsexplainer

Fine-tuning & adaptation3 lessonsupcoming

·Prompting vs RAG vs fine-tuning: the real decision treeexplain-back

·LoRA/QLoRA hands-on: dataset prep → train → comparecode-along

·Serving adapters & evaluating a fine-tune honestlycase-study

Multimodal & voice3 lessonsupcoming

·Vision inputs: documents, screenshots, image understandingcode-along

·Voice agents: STT/TTS pipelines vs realtime speech-to-speechexplainer

·Image & video generation APIs in productscase-study

Capstone & Career

AI system design & interviews4 lessonsupcoming

·AI system design: whiteboard a RAG/agent systemcase-study

·Python & SQL rapid revision for AI interviewscode-along

·Mock interview: defend your architecture choicesexplain-back

·Resume, GitHub portfolio & the story of your 9 projectscase-study

Capstone3 lessonsupcoming

·Capstone design: scope, architecture doc, eval planexplainer

·Build sprint: agentic RAG + MCP tools + guardrailscode-along

·Final eval, cost analysis, demo & self-reviewcase-study