NotebookLM Topic Organization Index

Derived from embeddings visualization — 35 top-level domains, 80+ skills, ~2,000+ document chunks across the MangaAssist knowledge base.

How to Use This Index

Each numbered section is a standalone NotebookLM notebook. Upload the files listed under each notebook into a single NotebookLM project. Notebooks are sized to stay within NotebookLM's 50-source limit.

Notebook 1 — AWS AIP-C01 Domain 1: FM Integration, Data & Compliance

Theme: Everything about integrating Foundation Models into a production system — architecture design, model selection, data pipelines, vector stores, retrieval, and prompt governance.

Folders / Files to Upload: - Domain1-FM-Integration-Data-Compliance/README.md - Domain1-FM-Integration-Data-Compliance/01-task-1.1-solution-design-and-requirements.md - Domain1-FM-Integration-Data-Compliance/02-task-1.2-foundation-model-selection-and-configuration.md - Domain1-FM-Integration-Data-Compliance/03-task-1.3-data-validation-and-processing-pipelines.md - Domain1-FM-Integration-Data-Compliance/04-task-1.4-vector-store-solutions.md - Domain1-FM-Integration-Data-Compliance/05-task-1.5-retrieval-mechanisms-for-fm-augmentation.md - Domain1-FM-Integration-Data-Compliance/06-task-1.6-prompt-engineering-and-governance.md - Domain1-FM-Integration-Data-Compliance/07-domain-1-scenario-map.md - All Skill-1.x.x-*/ scenario and interview Q&A files

Key Skill Groups: | Task | Skills | Focus | |------|--------|-------| | 1.1 — Solution Design | 1.1.1, 1.1.2, 1.1.3 | Architecture, PoC, Standardized Components | | 1.2 — FM Selection | 1.2.1, 1.2.2, 1.2.3, 1.2.4 | Assessment, Dynamic Selection, Resilience, Customization | | 1.3 — Data Pipelines | 1.3.1, 1.3.2, 1.3.3, 1.3.4 | Validation, Multimodal, Formatting, Quality Enhancement | | 1.4 — Vector Stores | 1.4.1, 1.4.2, 1.4.3, 1.4.4, 1.4.5 | Architecture, Metadata, High-Perf Search, Integration, Maintenance | | 1.5 — Retrieval | 1.5.1–1.5.6 | Segmentation, Embeddings, Search, Advanced Architectures, Query Handling | | 1.6 — Prompt & QA | 1.6.1–1.6.5 | Instruction Frameworks, Interactive AI, Governance, QA, Performance |

Notebook 2 — AWS AIP-C01 Domain 2: AI Implementation & Integration

Theme: Building and deploying intelligent, agentic AI systems — autonomous agents, tool use, deployment patterns, secure access, real-time interaction, and advanced GenAI apps.

Folders / Files to Upload: - All files under Implementation-Integration-Domain2/

Key Skill Groups: | Group | Skills | Focus | |-------|--------|-------| | 2.1 — Agentic AI | 2.1.1–2.1.7 | Autonomous Systems, Problem Solving, Safeguards, Model Coordination, Collaborative AI, Tool Integrations, Extensions | | 2.2 — FM Deployment | 2.2.1, 2.2.3 | Deployment Patterns, Optimized Deployment | | 2.3 — Enterprise Connectivity | 2.3.1, 2.3.2, 2.3.3 | Connectivity, Integrated AI, Secure Access / RBAC | | 2.4 — Model Interaction | 2.4.1–2.4.4 | Flexible Interaction, Real-Time/Streaming, Resilient Systems, Intelligent Routing | | 2.5 — Interfaces & Apps | 2.5.1–2.5.6 | FM APIs, Accessible Interfaces, Business Systems, Developer Productivity, Advanced Apps, Troubleshooting |

Notebook 3 — AWS AIP-C01 Domain 4: Operations, Monitoring & Optimization

Theme: Running GenAI in production — token efficiency, cost-effective model selection, caching, throughput, responsive AI, retrieval performance, and full-stack observability.

Folders / Files to Upload: - All files under Operational-Efficiency-Optimization/ - All files under Monitoring-GenAI-Systems/

Key Skill Groups: | Group | Skills | Focus | |-------|--------|-------| | 4.1 — Cost & Tokens | 4.1.1–4.1.4 | Token Efficiency, Cost-Effective Model Selection, High-Perf FM Systems, Intelligent Caching | | 4.2 — Performance | 4.2.1–4.2.6 | Responsive AI, Retrieval Performance, Throughput, FM Enhancement, Resource Allocation, System Performance | | 4.3 — Observability | 4.3.1–4.3.6 | Holistic Observability, GenAI Monitoring, Integrated Observability, Tool Performance, Vector Store Ops, FM Troubleshooting |

Notebook 4 — Security, Safety & AI Governance

Theme: Defending GenAI applications — prompt injection, PII protection, guardrails pipelines, content moderation, incident response, ML-specific threats, supply chain risk, encryption, and responsible AI principles.

Folders / Files to Upload: - All files under Security-Privacy-Guardrails/ - All files under AI-Safety-Security-Governance/

Key Topic Groups: | Area | Content | |------|---------| | Input/Output Safety | Prompt injection defense, guardrails pipeline deep dive, multi-turn context poisoning | | PII & Privacy | PII protection, data privacy controls (Skill 02) | | Content Moderation | Abuse prevention, system prompt extraction (Skill 04) | | Incident Response | Forensics, runbooks (Skill 05) | | ML-Specific Threats | Adversarial AI, model inversion, membership inference (Skill 06) | | Supply Chain & Encryption | Third-party risk, key management (Skills 07, 08) | | Governance & Responsible AI | Compliance frameworks, responsible AI principles (Skills 03, 04 of AI-Safety-Security-Governance) | | Storytelling & Interview | Security interview scenarios, STAR-D narratives (Skills 09, 10) |

Notebook 5 — Evaluation Systems & Testing

Theme: How to evaluate, test, and continuously improve a GenAI system — FM output quality, model comparison, user-centered feedback, offline testing strategies, edge cases, and prompt optimization.

Folders / Files to Upload: - All files under Evaluation-Systems-GenAI/ - All files under Offline-Testing-Quality-Strategies/ - All files under Cost-Optimization-Offline-Testing/

Key Topic Groups: | Area | Content | |------|---------| | FM Output Quality | RAGAS metrics, LLM-as-judge, faithfulness scoring (Skill 01) | | Model Evaluation | Sonnet vs Haiku benchmarking, optimal config (Skill 02) | | User-Centered Evaluation | Thumbs feedback interface, annotation workflows, feedback-to-action pipeline (Skill 03) | | Offline Testing | E2E integration testing, edge case playbook, prompt optimization workflow | | Cost-Efficient Testing | Change-type matrix, cheap testing strategy, regression prevention |

Notebook 6 — Statistical Foundations & Inference

Theme: The statistical underpinning of decisions made in GenAI systems — confidence intervals, t-tests, A/B testing, distribution shift detection, and monitoring statistics.

Folders / Files to Upload: - All files under Statistical-Inference/ - All files under Statistics-Foundations/ - model_evaluation_framework_deep_dive.md

Key Topics: - Confidence intervals for model metrics - T-tests for A/B and canary deployments - Pearson/Spearman correlations in feedback analysis - Golden dataset evaluation layers - Distribution shift detection (chi-squared, MMD, Mahalanobis) - Tools and libraries for statistical monitoring

Notebook 7 — Fine-Tuning, Model Customization & LLMOps

Theme: Adapting foundation models for a specific domain — LoRA/QLoRA, continual learning, RLHF, RAFT, prefix tuning, data curation, synthetic generation, and managing the full model lifecycle.

Folders / Files to Upload: - All files under Fine-Tuning-Foundational-Models/ - All files under LLMOps/ - All files under Model-Inference/

Key Topic Groups: | Area | Content | |------|---------| | PEFT Methods | LoRA, QLoRA, prefix tuning, prompt tuning | | Training Strategies | Continual learning, catastrophic forgetting, RLHF | | Advanced Fine-Tuning | RAFT, distillation, multi-task fine-tuning, data curation | | Synthetic Data | Self-instruct, active learning, distribution shift detection | | Model Inference | GPU architecture, KV cache, batching, quantization | | LLMOps Lifecycle | Model versioning, rollout, A/B testing, lineage, deprecation |

Notebook 8 — Prompt Engineering

Theme: Designing, optimizing, and governing prompts at scale — chain-of-thought, few-shot, role prompting, failure scenarios, and system prompt management.

Folders / Files to Upload: - All files under Prompt-Engineering/

Key Topics: - Foundational prompt design techniques - Chain-of-thought and step-back prompting - Foundational model optimization techniques - Failure scenarios and workarounds (hallucinations, price errors) - Interview-prep drills for prompt engineering

Notebook 9 — RAG Pipeline & Vector Search

Theme: The full retrieval-augmented generation pipeline — chunking, embedding, hybrid search, MCP integrations, query handling, and architecture patterns for production RAG.

Folders / Files to Upload: - All files under RAG-MCP-Integration/ - Relevant Domain1-FM-Integration-Data-Compliance/Skill-1.4.x and Skill-1.5.x files (cross-reference with Notebook 1)

Key Topics: | File | Topic | |------|-------| | 00-overview-rag-mcp-architecture.md | End-to-end RAG + MCP architecture overview | | 01-catalog-search-mcp.md | Catalog search via MCP tool | | 02-user-preferences-recommendation-mcp.md | Personalized recommendations | | 03-order-inventory-mcp.md | Order and inventory lookups | | 04-review-sentiment-mcp.md | Review sentiment MCP pipeline | | 05-support-policy-mcp.md | Support and policy RAG | | 06-trending-discovery-mcp.md | Trending content discovery | | 07-cross-title-link-mcp.md | Cross-title/related item linking | | 08-mcp-orchestration-router.md | MCP orchestration and routing | | 09-rag-retrieval-pipeline-deep-dive.md | Deep dive — chunking, reranking, hybrid search | | 10-mcp-basics-and-amazon-backend-integration.md | MCP basics and backend integration |

Notebook 10 — AWS Infrastructure & Data Services

Theme: The AWS services that power MangaAssist — DynamoDB data modelling, ECS/Fargate/Lambda compute, Docker containers for ML, and the overall tech stack including vLLM.

Folders / Files to Upload: - All files under DynamoDB/ - All files under ECS-Fargate-Lambda/ - All files under Docker-Interview-Pack/ - All files under Tech-Stack/ - All files under MLflow/

Key Topic Groups: | Service / Tool | Area | |----------------|------| | DynamoDB | Data modelling, access patterns, streams, pipelines, security, observability | | ECS Fargate | Container orchestration, task definitions, auto-scaling | | Lambda | Event-driven patterns, cold starts, triggers | | Docker | Containerizing ML workloads, vLLM serving containers | | vLLM | Continuous batching, paged attention, throughput optimization | | MLflow | LLM observability, experiment tracking, tracing |

Notebook 11 — System Design: HLD & LLD

Theme: Designing and low-level implementing the MangaAssist system — high-level architecture questions, low-level design walkthroughs, and API design strategy.

Folders / Files to Upload: - All files under HLD-Questions/ - All files under LLD-Questions/ - All files under API-Design-and-Testing/

Key Topic Groups: | Area | Content | |------|---------| | HLD | RAG pipeline, chatbot scalability, recommendation engine, data ingestion, caching | | LLD | Component-level design — session management, embedding pipeline, guardrails | | API Design | REST vs WebSocket, API testing strategy, scale testing, offline quality | | Grilling Sessions | Hard follow-up drills, system design under pressure |

Notebook 12 — CI/CD Pipelines & DevOps

Theme: Shipping ML systems reliably — deployment pipelines, model promotion pipelines, database migrations, canary deployments, and behavioural tests.

Folders / Files to Upload: - All files under CI-CD-Pipeline-User-Stories/

Key Topics: - CD-01: Application code deployment pipeline - CD-02–06: Model pipeline, data pipeline, infrastructure, rollback - CD-07: Database migration pipeline - Behavioural tests for GenAI deployments

Notebook 13 — Cost & Performance Optimization User Stories

Theme: Product-level tradeoffs and optimization decisions — LLM model tiering, caching policies, token budgets, cost dashboards, and unified optimization decisions.

Folders / Files to Upload: - All files under Cost-Optimization-User-Stories/ - All files under Performance-Optimization-User-Stories/ - All files under Optimization-Tradeoffs-User-Stories/

Key Topics: - LLM model tiering (quality vs cost vs latency tradeoffs) - Caching strategy user stories - Token budget management - Performance SLA user stories - Unified optimization decision dashboard

Notebook 14 — Troubleshooting & Debugging GenAI Applications

Theme: Diagnosing and resolving failures in production GenAI systems — application logging, distributed tracing, POC failure patterns, and systematic debugging runbooks.

Folders / Files to Upload: - All files under Troubleshoot-GenAI-Applications/ - All files under Debugging/

Key Topics: - Application logging across a multi-service chatbot - End-to-end turn debugging - POC implementation pitfalls and fixes - Prompt observability pipeline - Prompt A/B testing framework debugging

Notebook 15 — Math & Modeling Foundations

Theme: The mathematical foundations underlying the ML models in MangaAssist — linear algebra, regression, neural network architectures, and embedding/retrieval math.

Folders / Files to Upload: - All files under Modeling-Math-Foundations/

Key Topics: | File | Content | |------|---------| | 01-matrix-theory-and-linear-algebra.md | Matrices, SVD, PCA for embeddings | | 02-linear-regression-and-generalized-models.md | Regression, logistic, GLMs | | 03-neural-network-architectures.md | Transformers, attention, MLP | | 04-embedding-and-retrieval-architectures.md | Bi-encoders, cross-encoders, HNSW | | 05-tools-and-libraries.md | NumPy, PyTorch, scikit-learn usage |

Notebook 16 — Interview Preparation & War Stories

Theme: Structured interview practice — architect-level questions, real-world war stories, behavioural questions, and the complete MangaAssist narrative.

Folders / Files to Upload: - All files under MangaAssist-Interview-Pack/ - All files under mangaassist_workflow_interview_pack/ - All files under Real-World-Interview-Questions/ - All files under Challenges/ - All files under POC-to-Production-War-Story/ - Root architecture files: 01-problem-statement.md through 16-final-summary.md

Key Topic Groups: | Area | Content | |------|---------| | MangaAssist Architecture Story | Problem statement → HLD → LLD → Metrics → Tradeoffs | | Architect-Level Questions | System design, scaling, reliability, failure modes | | Real-World Challenges | 20+ production challenges at scale | | POC-to-Production | The PoC that fooled us — lessons from production failure | | Workflow Pack | End-to-end chatbot workflow interview scenarios |

Quick Reference: Domain → Notebook Mapping