Concrete staircase and steel rail in a modern atrium

Practical writing for people who have to live with the system.

Articles written to support real service conversations — migration, delivery, architecture, data, and CRM.

Featured · AI EngineeringTaking an LLM feature from demo to production: what changes after the prototypeA prototype proves a model can do the task once. Production needs a definition of good, an evaluation set, grounded data, guardrails, and someone watching cost and quality.Read article →
Cloud & InfrastructureCloud migration planning: the risks that appear after the kickoffMost migrations fail in the unglamorous middle: incomplete inventory, identity left until last, and cutovers that cannot be reversed.DevOps & ReliabilityCI/CD pipeline design that a team will still understand in a yearPipelines should produce an artefact you can promote. Everything else is ceremony that needs a reason.Software ArchitectureMonolith versus microservices is the wrong first questionStart from team shape, data ownership, and release pain. The topology should follow those facts.Data EngineeringData pipeline reliability is a production problemA dashboard that looks fine on stale data is a quieter kind of outage. Pipelines need the same seriousness as user-facing services.Data EngineeringWarehouse or lakehouse: choose the constraint, not the labelThe useful choice is about file types, freshness, and who will operate the platform — not about winning an architecture argument.Analytics & Decision IntelligenceDesigning business intelligence people will keep openStart from the meeting the numbers must survive. Then write definitions. The tiles come last.CRM & Enterprise SystemsCRM integration and data migration without a second source of truthA CRM should not invent a second customer. Migration is a design problem before it is an import problem.Software ArchitectureModernizing a legacy application without a heroic rewriteThe system that makes money is allowed to be imperfect. Your job is to make the next change cheaper than the last.Cloud & InfrastructureCloud cost governance without a theatre of dashboardsCost is an architecture and ownership problem. A weekly screenshot of the invoice is not FinOps.DevOps & ReliabilityObservability and production readiness are the same projectIf you cannot tell that a customer is stuck, you are not ready — regardless of how many metrics you store.