Training, storing, serving and governing models — classical and generative — including where tenant data may go.
Work that happens outside a request: queues, workers, scheduled jobs, batch orchestration, event streams.
Language, runtime, framework, API style, request validation, and how business logic is organised.
Pipeline, environments, database migration strategy, and how releases reach production without breaking anything mid-flight.
Where per-tenant configuration lives, how it is versioned, promoted between environments and tested. Distinct from application config.
The primary datastore and, more consequentially, how records are shaped: mutability, temporality, numeric precision, and history.
Ingesting, parsing, generating and storing files: uploads, format parsing, extraction, export generation.
How the user interface is built, rendered and delivered: framework, rendering strategy, state management, component system, build tooling.
Who a user is, what they may do, and how rules more complex than roles are expressed and enforced.
Compute, infrastructure-as-code, networking, backup, disaster recovery and cost management.
Moving data in and out of third-party systems: protocols, file transfer, API clients, retry and reconciliation of transfer itself.
How a developer runs and tests the system on their own machine, especially when it integrates with systems they cannot run locally.
How tenants are separated in storage, compute and configuration, and what guarantees that separation.
How the system tells people something needs their attention: channels, delivery, escalation, digest versus immediate.
Logging, metrics, tracing, alerting and cost visibility — knowing what the system is doing and what it is costing.
Partitioning, indexing, caching, read replicas, and the shape of load the system must absorb.
Serving aggregate queries and exports without degrading the transactional system.
Finding records across large volumes: full-text, faceted, and increasingly semantic search.
Encryption, secrets, audit logging, data residency, retention periods and the certifications customers require.
Unit, integration, contract, load and — for data-processing systems — correctness and data-quality testing.
Representing instants, dates, periods, business days and holiday calendars correctly across jurisdictions.
How multi-step, long-running, human-in-the-loop processes are modelled, executed, resumed and audited.