AI Engineering¶

The AI-engineering track is organized by module, not by content type. Each module folder contains the topic files for one pressure area, with optional hands_on_lab/ and extended_notes/ folders when a module needs exercises or archived deep dives.

The track is grouped into six phases, and the default learning motion is agent-first: frame the product problem, build one useful agent, then pull in the surrounding systems only when the agent exposes the need for them.

Reading order note: Do not start by grinding all foundations. If you need model mechanics, use ../00_ai_foundation/. If you need serving, vector infrastructure, MLOps, or cost/latency mechanics, use ../02_ai_infrastructure/. If you need safety/security or evals, use ../03_ai_security_safety/ and ../04_ai_product_evals/. Then return here and keep moving from agents outward.

Phase 1 — Product framing and agent core¶

Module	Focus	Folder
00	AI product requirements — constraints, success criteria, acceptance tests	`00_ai_product_requirements/` (placeholder)
01	Agentic system design — architecture, tools, blast radius, rollout	`01_agentic_system_design/`
02	Durable agent workflows — orchestration, state, retries, recovery	`02_durable_agent_workflows/`
03	Agent observability and debugging — taxonomy, traces, regression locks	`03_agent_observability_debugging/`

Phase 2 — Agent operations¶

Module	Focus	Folder
04	Resilient agent systems — fallbacks, retries, degradation	`04_resilient_agent_systems/`
05	AI incident operations — sev levels, rollback, postmortems	`05_ai_incident_operations/`

Phase 3 — Context, retrieval, and memory¶

Module	Focus	Folder
06	Evidence and data pipelines for AI	`06_evidence_data_pipelines/`
07	Search relevance and ranking	`07_search_relevance_ranking/`
08	RAG system design	`08_rag_system_design/`
09	Advanced RAG patterns	`09_advanced_rag_patterns/`
10	Knowledge-graph retrieval	`10_knowledge_graph_retrieval/`
11	Long-term memory and state	`11_long_term_memory_state/`

Phase 4 — Model, prompt, and migration strategy¶

Module	Focus	Folder
12	Model and vendor strategy	`12_model_vendor_strategy/` (placeholder)
13	Prompt lifecycle operations — prompts as code	`13_prompt_lifecycle_operations/` (placeholder)
14	Legacy AI modernization — inheriting and stabilizing v1 systems	`14_legacy_ai_modernization/` (placeholder)

Phase 5 — Reasoning, multi-agent systems, and generated artifacts¶

Module	Focus	Folder
15	Reasoning, routing, and verification	`15_reasoning_routing_verification/`
16	Multi-agent coordination and protocols	`16_multi_agent_coordination/`
17	Schema-driven generation	`17_schema_driven_generation/`

Phase 6 — Product experience and tool contracts¶

Module	Focus	Folder
18	Human-AI product experience	`18_human_ai_product_experience/` (placeholder)
19	Tool integration contracts	`19_tool_integration_contracts/` (placeholder)

Phase 7 — Leadership and capstone¶

Module	Focus	Folder
20	Engineering leadership judgment	`20_engineering_leadership_judgment/`
21	Capstone agentic AI system	`21_capstone_agentic_ai_system/`

Phase 8 — Platform strategy and engineering leverage¶

Module	Focus	Folder
22	Agentic platform evaluation — build-vs-buy, hyperscaler/AI-native/SaaS landscape, capability rubric, lock-in, cost, governance	`22_agentic_platform_evaluation/`
23	GenAI for the SDLC — coding assistants, spec-to-code, AI review, test/doc generation, ops copilots, productivity measurement, IP/security	`23_genai_for_sdlc/`

This is the primary content track for AI-specific learning. Use ../06_system_designing/ as the parallel track for architecture and interview-oriented systems thinking.

For prioritization from a lead-AI-engineer interview and production-ownership perspective, see lead_ai_engineer_module_priority.md. For the rewrite backlog, see module_rewrite_tasks.md.