Concepts

Agent planning (ReAct, Reflexion); tool categories (knowledge, capability, write actions); function calling; multi-agent systems; planning/tool/efficiency failure modes; safety and compound errors; LLM planning debate

AI engineering as a discipline vs ML engineering; three-layer stack (application dev, model dev, infrastructure); product planning; last-mile challenge; use case taxonomy

Five-step progressive architecture (context enhancement, guardrails, router/gateway, caching, agents); model drift detection; orchestration frameworks; user feedback systems (explicit vs implicit, edit-as-preference, degenerate feedback loops, sycophancy)

Evaluation methodology: perplexity, pass@k, BLEU/ROUGE, embeddings, AI-as-a-judge, comparative evaluation (Elo/Bradley-Terry, LMSYS Arena); factual consistency, safety, instruction-following; model selection workflow; API vs self-hosting; evaluation pipeline design; benchmark contamination

Data Warehouse, Data Lake, Data Mesh evolution; four Data Mesh principles; DPQ; OLTP vs OLAP; star/snowflake schema; DDD/Data Mesh alignment (OHS as analytical published language, CQRS for projections)

REST, gRPC, GraphQL exchange format decision; HTTP methods; URL modeling; status codes; OAS; versioning; Postel's Robustness Principle; non-breaking vs breaking changes; implementation-as-de-facto-spec; version translation in controllers; API-first

Gateway taxonomy (enterprise/microservices/mesh), capabilities, pitfalls, history

Test pyramid, consumer-driven contracts, Pact, component/integration/E2E testing; scope vs size distinction, test double fidelity hierarchy, user journey tests, formal verification with TLA+; EDM topology testing, integration testing strategies, event data sourcing

Diagramming, presenting, team leadership (architect personality types, elastic leadership, team warning signs, checklists), negotiation (4 C's, stakeholder/architect/developer tactics), career (20-minute rule, personal technology radar)

Modularity drivers (availability, scalability, deployability, testability, maintainability); scalability vs elasticity; component-based vs tactical forking; six decomposition patterns; architecture stories

The "-ilities": taxonomy, selection, measurement, governance

Architecture Decision Records: structure, anti-patterns, compliance governance, Type 1/2 decisions

Purpose and audience taxonomy; C4 / UML / ArchiMate notations; ADRs as the durable documentation form; diagrams-as-code; "just sufficient" principle; presentation craft (two-channel model, infodecks vs presentations, Bullet-Riddled Corpse anti-pattern)

Full classification taxonomy; priority tiers; enterprise fitness functions; mechanism and concern taxonomies; fitness function testing pyramid (triggered/continuous × atomic/holistic → three pyramid layers); ISO 25010 quality attribute anchoring

Independently deployable artifact; static/dynamic coupling distinction; quantum count by architecture style; micro-frontends (including EDM pairing); 3D dynamic coupling space → 8 saga types

DDD strategic design: subdomain taxonomy (core/generic/supporting), bounded context definition, subdomains vs bounded contexts, physical/ownership boundaries, sizing heuristic (start wide), subdomain type evolution (all 6 transitions), growth management; Evans' originating treatment: Continuous Integration within context, false cognates vs duplicate concepts, context sizing trade-offs

Four reuse techniques: code replication, shared library, shared service, sidecar; core reuse principle (reuse = abstraction + slow rate of change); versioning discipline; orthogonal coupling

Stateless retraining vs stateful training (fine-tuning); four stages of continual learning maturity; champion/challenger pattern; test in production methods (shadow deployment, A/B testing, canary release, interleaving experiments, bandits, contextual bandits); data freshness value; continuous ML challenges (feedback loops, emergency response, stable A/B baselines)

Eight Accelerate-validated CD capabilities; key empirical findings: config-in-VCS more predictive than code, developer-owned tests, trunk-based development, loosely coupled architecture as the single largest CD lever; shift-left security (DevSecOps); lightweight change approval vs CABs; relationship between CD, culture, burnout, and job satisfaction

Strict/loose contract spectrum; consumer-driven contracts; stamp coupling; event schema evolution (forward/backward/full compatibility); schema registry; event design anti-patterns; Newman: structural vs semantic breakages, expansion changes, dual-version strategies; EIP: Format Indicator (version number/foreign key/embedded schema), Canonical Data Model

Conway's Law; Inverse Conway Maneuver; three organisational structures (Pflaeging); tool choices as communication drivers; unexpected communication as diagnostic signal; organisation design as technical work

strategic distillation: CORE DOMAIN identification, generic subdomains, domain vision statement, highlighted core, cohesive mechanisms, segregated core, abstract core; escalation of distillation techniques

Cost as first-class quality attribute; five cost dimensions (build/run/change/downtime/opportunity); architecture style as dominant cost driver; ROI patterns (deployment automation 18×, observability, stability patterns); common cost traps; cost-vs-other-ilities trade-offs

Coupling taxonomy: structural (Ca/Ce, Martin A/I/D, connascence), operational (Newman's implementation/temporal/deployment/domain types), contract (strict/loose, stamp coupling), integration-style spectrum (File Transfer → Messaging); coupling and quantum boundaries; coordinated deployments as coupling signal; concept leakage and flatten-when-publishing

Data disintegrators/integrators; five-step process; data domains; data sovereignty; connection quotas; polyglot persistence; database type selection; data ownership (single/common/joint); eventual consistency patterns; four distributed data access patterns

Covariate shift/label shift/concept drift/feature change/label schema change taxonomy; degenerate feedback loops (mechanism, detection, correction via randomisation and positional features); detection methods (KS test, two-sample tests, feature validation, prediction monitoring, sliding vs cumulative statistics); proactive design; retraining strategies

Data-centric AI; three criteria (quality, coverage, quantity); acquisition and annotation; data synthesis (rule-based, simulation, AI-powered, reverse instruction); model distillation; model collapse; data processing pipeline

Fitness function automation; stages, fan-in/fan-out; CD vs CDP; cycle time as business metric; zero-downtime deployment (four phases: prepare/drain/apply/start); expand/contract relational schema pattern; trickle-then-batch schemaless migration; canary group evaluation; immutable infrastructure; EDM deployment patterns

Low-tech collaborative workshop; 10-step process (domain events, commands, policies, aggregates, bounded context candidates); when to use; two-phase facilitation; remote considerations; Rosa: Big Picture EventStorming for domain boundary/KPI mapping + KPI Value Tree; Process Modeling EventStorming for operational value stream analysis

Three pillars; evolvability by style; strangler fig; six Rs; guidelines; antipatterns; DDD brownfield modernisation strategy; "form follows failure"; bad layering (horizontal coupling); component-based decomposition; decision loop speed and thrashing; service extinction; Farley: testability+deployability as operational drivers, TDD as architectural feedback, architectural descriptions as "tourist maps"

Expand/contract pattern; Flyway/Liquibase; shared DB decomposition; reporting antipattern; Newman's ~12 database decomposition patterns (database view, wrapping service, tracer write, synchronize in application, etc.)

Feature operations (missing value handling MNAR/MAR/MCAR, scaling, discretisation, categorical encoding/hashing trick, feature crossing, positional embeddings); data leakage causes and detection; feature importance (top 10 = 50% importance); feature generalisation (coverage × distribution overlap); best practices (split by time, scale after split)

Application-layer release control; Hodgson's four toggle types (release/experiment/ops/permission) with different lifecycles; flag lifecycle discipline; default behaviour and cohort routing; kill switches and load-shedding flags; flags as technical debt with a use-by date

When to finetune vs RAG; memory bottleneck; quantisation (FP32→INT4); PEFT; LoRA (low-rank decomposition, rank, alpha, multi-LoRA serving); QLoRA; model merging (task arithmetic, TIES/DARE, frankenmerging); finetuning tactics

What foundation models are; transformer architecture (prefill/decode, KV cache, MoE); training data constraints; Chinchilla scaling law; post-training (SFT, RLHF, DPO)

DORA four key metrics (deployment frequency, lead time for changes, change failure rate, time to restore service); empirical basis (4-year, 23K+ respondents, Accelerate); Lean conceptual basis; cluster analysis; no-tradeoff finding; CABs negatively correlated with both metrics; Goodhart's Law warning; pipeline topology variants; instrumentation; MVD; virtuous cycle mechanism; complementary metrics

Eight types of natural split points for monolith decomposition: business domain (primary), regulatory compliance, change cadence, team location, risk, performance, technology (sparingly), user personas; six forms of monolith; distributed monolith anti-pattern

GQM framework (Basili & Weiss 1984): goal → questions → metrics → data tree; goal statement structure (purpose/object/issue/viewpoint); metric selection criteria (signal strength, cost, cross-question reuse); 9-step workshop; Foo Service case study (rate-limit incident → heartbeat + fail-fast ADR → 10-minute detection before user impact)

Inference metrics (TTFT, TPOT, throughput, goodput, MFU, MBU); prefill (compute-bound) vs decode (memory bandwidth-bound); speculative decoding; KV cache management (PagedAttention, FlashAttention, GQA/MQA); continuous batching; prefill-decode decoupling; prompt caching

the four EIP integration styles (File Transfer, Shared Database, RPI, Messaging); eight decision criteria; trade-offs and when to use which style

organising principle for entire system; four patterns: System Metaphor, Responsibility Layers (Potential/Operations/Decision Support/Policy/Commitment), Knowledge Level, Pluggable Component Framework; Evolving Order; six essentials for strategic design

Sampling strategies (temperature, top-k, top-p); test-time compute; structured outputs; hallucination (self-delusion + knowledge-mismatch hypotheses); inconsistency mitigation

Command/Document/Event message types; one-way/req-resp/broadcast styles; Request-Reply (sync block vs async callback; Return Address; Correlation Identifier); point-to-point/pub-sub/datatype channels; Pub-Sub as distributed Observer (push vs pull; channel design); at-least-once delivery; guaranteed delivery; exactly-once via idempotency; invalid vs dead letter channel; backlogs; poison message isolation; competing consumers; data safety trade-off (publisher confirms + persistent queues + manual ACKs); quorum queues (RAFT); RabbitMQ internals; Pipes and Filters; Message Router; Message Bus; Messaging Bridge; Aggregator; Resequencer; Composed Message Processor; Scatter-Gather; Routing Slip; Process Manager; Message Broker (architectural pattern); Envelope Wrapper; Content Enricher; Content Filter; Claim Check; Normalizer; Canonical Data Model; Messaging Gateway; Messaging Mapper; Transactional Client; Polling Consumer; Event-Driven Consumer; Competing Consumers; Message Dispatcher; Selective Consumer; Durable Subscriber; Idempotent Receiver; Service Activator; Control Bus; Detour; Wire Tap; Message History; Message Store; Smart Proxy; Test Message; Channel Purger

When to use ML (nine conditions); research vs production differences (silent failure, latency vs throughput, messy data, fairness, interpretability); four system requirements (reliability, scalability, maintainability, adaptability); business vs ML objective alignment; problem framing (task types, decoupling objectives); mind vs data debate

Four-layer MLOps stack (storage/compute, resource management, ML platform, dev environment); cloud repatriation; Docker + Kubernetes; workflow orchestration tools (Airflow/Argo/Prefect/Metaflow); model store (8 artifact types); feature store (management/computation/consistency, eliminates training-serving skew); build vs buy framework

Six model selection tips; four phases of ML adoption; ensembles (bagging/boosting/stacking); experiment tracking and data versioning challenges; ML debugging; distributed training (data/model/pipeline parallelism); AutoML (hyperparameter tuning, NAS, learned optimisers); offline evaluation baselines; evaluation methods (perturbation, invariance, directional expectation, calibration, confidence, slice-based/Simpson's paradox)

Evans' binding principle: one model for analysis and design; code is the model's expression; hands-on modellers; bones-showing; requires layered architecture for domain isolation; breakthroughs: non-linear returns, cascade of insights, Share Pie story

Cohesion types, LCOM, coupling metrics, connascence taxonomy (static vs dynamic); SATH: applying D/A/I metrics to decomposition feasibility; Modularity Maturity Index (MMI 0-10, three cognitive principles: modularity 45%/hierarchy 30%/pattern consistency 25%, decision thresholds, architecture erosion, implementation vs design/architecture debt); advanced structural metrics: ACD/CCD/Propagation Cost, Relative Cyclicity, SDI, Maintainability Level, LCOM4, change history metrics, Component Rank, six golden rules

OAuth2 roles, JWT, grants (Auth Code, PKCE, Client Credentials), OIDC, refresh tokens

Prompt anatomy; system vs user prompt; chat templates; in-context learning (zero/few-shot, CoT); prompt decomposition; defensive prompt engineering (extraction, jailbreaking, injection, information extraction); three-layer defence (model/prompt/system)

Retrieval-augmented generation; term-based (TF-IDF, BM25) vs embedding-based retrieval; ANN algorithms (LSH, HNSW, FAISS); hybrid search with RRF; chunking; reranking; contextual retrieval; text-to-SQL; memory model

Collaborative risk identification: risk matrix, risk assessment, 3-activity process

Granularity vs modularity distinction; six disintegrators; four integrators; volatility-based decomposition; MTTS; trade-off-to-business-question method

Ousterhout's complexity definition: three symptoms (change amplification, cognitive load, unknown unknowns), two causes (dependencies, obscurity), incremental accumulation; strategic vs. tactical programming; technical debt as borrowed time

Coherent navigation hub for Nygard's stability pattern set; failure-arresting (timeout/circuit-breaker/bulkhead), load-handling (shed load/backpressure/handshaking), defensive (fail fast/let it crash/steady state/retry), and architectural (decoupling middleware/test harnesses/governor); pattern composition; antipatterns they defend against

Evans' six patterns for design that is easy to work with and extend: Intention-Revealing Interfaces, Side-Effect-Free Functions, Assertions, Conceptual Contours, Standalone Classes, Closure of Operations; declarative style as outcome

Sweller's three cognitive load types (intrinsic/extraneous/germane); team cognitive capacity as software boundary constraint; domain complexity heuristics; platform as extraneous load eliminator

Four team types (stream-aligned, platform, enabling, complicated-subsystem); three interaction modes with constraints/matrix; Dunbar-based team sizing; team API; one-owner-per-component rule; promise theory / SemVer; organisational sensing; discovery-to-establish pattern; topology evolution triggers; TVP; fractal platforms; Spotify model; DevOps anti-patterns

Technical (layered) vs domain (bounded context) top-level component design

STRIDE, DREAD, OWASP API Top 10, six-step process, rate limiting strategies

DDD cornerstone practice: shared business language eliminating translation chains; must pervade the code; modeling out loud; documents must track UL or be archived; UML limitations; explanatory models as separate teaching aids

Ron Westrum's organisational culture typology (pathological/bureaucratic/generative); information flow as core mechanism; Likert measurement with psychometric validation; predicts delivery performance, org performance, job satisfaction; how to improve culture (implement practices first); blameless postmortems; Google Project Aristotle alignment

Zonal architecture critique, NCSC eight principles, service mesh + OAuth2 implementation stack