Review menu

Every chapter, slide deck and quiz set has its own issue. Pick the ones that match what you know. Each issue is deliberately open-ended: read, then tell us what you find.

On this page
  1. How it works
  2. Start here
  3. By topic
    1. Agent design
    2. Reasoning, planning and memory
    3. Retrieval and data
    4. Evaluation and improvement
    5. Deployment and operations
    6. Safety and responsibility
    7. Humans and agents
  4. By Part
  5. What kinds of review help

How it works

  1. Find an issue below (or browse the open review issues).
  2. Comment on it to say you’d like to take it, and which angle you plan to take if you have one. A maintainer assigns it to you.
  3. For a chapter, we email you the chapter text. Chapter text is not published in this repository. Slide decks and quizzes are already public, so you can start right away.
  4. Post what you find as comments on the issue: where it is, what you found, a quote or source, and a suggested fix. The review checklist says what a useful finding looks like.
  5. Say where you land: accept, minor revisions, or major revisions. A partial review (one section, or one angle) is a real contribution.

There is no minimum and no fixed list to work through. Contributors are credited through All Contributors.

Start here

I want… Open issues
Anything nobody has taken yet Unassigned reviews
A small first review good first issue
Something short (about 1 to 3 hours) effort:small
A real project (3 to 8 hours) effort:medium
A deep read (8+ hours) effort:large
Chapter text artifact:chapter-text
A slide deck, no chapter text needed artifact:slides
Quizzes, which are public Google Forms artifact:quiz
To see progress Part trackers

Filters combine. For example, unassigned, small chapter reviews.

By topic

Topics cut across Parts, so you can follow one subject through the whole book. Each chapter has one to three topic labels, the main one first. The chapter numbers below link to the chapter’s issue.

Agent design

Topic Covers Chapters  
topic:agent-patterns Core agent patterns: ReAct, plan-and-execute, reflection, hybrid designs 1.1A, 1.2, 2.1, 5.13, 10.2 open issues
topic:multi-agent Multi-agent collaboration, communication and coordination 1.3, 2.4 open issues
topic:orchestration-state Stateful orchestration, workflows, checkpointing, control flow 1.3, 1.5A, 1.5B, 1.6, 2.2 open issues
topic:frameworks LangChain, LangGraph, AutoGen, CrewAI, Semantic Kernel; framework selection 1.5B, 2.1, 2.2, 2.3, 2.4, 2.5, 2.6 open issues
topic:tool-use Tool integration, function calling, tool auditing and constraints 1.2, 2.3, 2.5, 2.6, 2.8, 3.3, 3.7, 3.8, 9.2 open issues

Reasoning, planning and memory

Topic Covers Chapters  
topic:reasoning Chain-of-thought, tree-of-thought, self-consistency, reasoning quality 1.2, 3.5, 3.9, 5.1, 5.2, 5.3 open issues
topic:planning-search Hierarchical planning, MCTS, A* and other search-based planning 5.2, 5.4, 5.5, 5.6, 6.6 open issues
topic:decision-making Utility-based, rule-based, learning-based and hybrid decision making 5.5, 5.10, 5.11, 5.12, 5.13 open issues
topic:memory Working, episodic, semantic and long-term memory for agents 1.4, 1.5A, 5.7, 5.8, 5.9 open issues
topic:knowledge-graphs Knowledge graphs, relational reasoning, hybrid RAG+KG 1.7A, 1.7B, 5.8 open issues

Retrieval and data

Topic Covers Chapters  
topic:rag-retrieval Embeddings, chunking, reranking, query decomposition, production RAG 1.7A, 1.7B, 2.7, 6.1, 6.2A, 6.5, 6.6 open issues
topic:data-pipelines Vector databases, ETL pipelines, data quality 6.2A, 6.2B, 6.3A, 6.3B, 6.4, 7.5 open issues
topic:multimodal Vision, audio and text fusion; perception and speech 1.4, 2.7, 7.5 open issues

Evaluation and improvement

Topic Covers Chapters  
topic:evaluation Benchmarks, evaluation pipelines and metrics 3.1A, 3.1B, 3.2, 3.3, 3.6, 3.8, 3.9, 3.10, 6.4, 8.4 open issues
topic:observability-debugging Tracing, execution debugging and auditing of agent behavior 3.6, 3.7, 4.4, 7.3 open issues
topic:tuning-optimization Prompt optimization, fine-tuning, parameter tuning, reward modeling 3.4, 3.5, 10.3 open issues

Deployment and operations

Topic Covers Chapters  
topic:deployment-scaling Containers, Kubernetes, edge, autoscaling, production deployment 1.8, 4.1, 4.2, 4.3, 4.5, 4.6, 4.7, 6.2B, 6.5, 7.6 open issues
topic:inference-performance Profiling, latency, quantization, GPU efficiency 3.4, 3.10, 4.4, 4.5, 4.6, 7.2, 7.4, 8.1 open issues
topic:nvidia-platform NIM, Triton, TensorRT-LLM, NeMo, Riva, Curator, MIG, Fleet Command 4.5, 4.6, 7.1A, 7.1B, 7.2, 7.3, 7.4, 7.5, 7.6, 8.2B open issues
topic:reliability-resilience Error handling, retries, circuit breakers, SLOs 1.6, 1.8, 2.8, 4.2, 8.2A, 8.2B open issues
topic:monitoring-metrics Production monitoring, latency, error-rate and success metrics 7.2, 8.1, 8.2A, 8.3, 8.4 open issues
topic:cost-economics Token economics, cost/performance trade-offs, efficiency 3.10, 4.7, 8.3 open issues
topic:streaming-realtime Streaming and real-time agent responses 2.9 open issues

Safety and responsibility

Topic Covers Chapters  
topic:guardrails-safety Output filtering, action constraints, sandboxing, guardrail frameworks 7.1A, 7.1B, 8.2B, 9.1, 9.2, 9.3, 9.5, 10.5 open issues
topic:fairness-alignment Fairness and bias, constitutional AI, value alignment 9.4, 9.5, 9.6 open issues
topic:governance-compliance GDPR, standards, certifications and AI governance frameworks 9.3, 9.6, 9.7, 9.8 open issues

Humans and agents

Topic Covers Chapters  
topic:human-ai-interaction UI design, conversational UI, accessibility, proactive agents 1.1A, 1.1B, 2.9, 10.1, 10.2 open issues
topic:human-oversight-feedback Human-in/over-the-loop, RLHF, feedback and red teaming 1.1B, 10.3, 10.4, 10.5 open issues

To follow two topics at once, separate the labels with a comma in the search box, for example label:"topic:memory","topic:knowledge-graphs" (matches either).

By Part

Part Title Open issues
1 Agent Fundamentals part:01
2 Framework & Tool Integration part:02
3 Evaluation & Optimization part:03
4 Production Deployment & Scaling part:04
5 Advanced Reasoning & Decision Making part:05
6 Retrieval-Augmented Generation (RAG) part:06
7 NVIDIA NeMo Framework & Optimization part:07
8 Reliability & Cost Management part:08
9 Safety & Governance part:09
10 Human-in-the-Loop & Integration part:10

What kinds of review help

  • Fact-checking. Are the claims right, and are tools, models, APIs and standards still current?
  • Expert reading. Does it match how practitioners actually build and operate these systems?
  • Learner’s eye. Is it clear, in a sensible order, and pitched at the right level? Where would a newcomer get lost?
  • Exam fit. Does it prepare a candidate for the certifications it is mapped to in cert_mapping/?
  • Consistency. Does it agree with neighboring chapters, the slides and the quizzes?

Questions? See SUPPORT, or ask on the issue.