Services
What CloudCraft does
Three ways CloudCraft works with teams building on AWS — design the system, drive down what it costs to run, or review what’s already in production. Each is scoped and delivered as something your team can act on.
GenAI architecture
Design and review of LLM and RAG applications on AWS. This is the system itself — Amazon Bedrock and model selection, retrieval pipelines, serverless backends (Lambda, API Gateway, DynamoDB, Aurora), and the multi-account infrastructure underneath. The goal is an architecture that does the job, holds up in production, and has its running cost designed in from the start rather than discovered later.
Cost & token-efficiency optimization
The specialty. A working GenAI system and an affordable one are not the same system, and the difference is architectural. CloudCraft works the levers that actually move the bill:
- Right-sizing model choice to the task instead of defaulting to the largest option.
- Context and token economics — controlling what goes into the context window.
- Prompt caching, so stable context isn’t paid for on every call.
- Batch vs. real-time routing for work that can wait.
- Retrieval cost and embedding strategy — an accurate RAG layer without over-retrieving.
- Model routing — small models for the easy requests, the expensive ones only where they earn it.
The target is the same quality for materially less money — not the cheapest system that technically runs.
GenAI architecture & cost reviews
A scoped assessment of an existing system. CloudCraft examines the architecture and the economics together and reports back: what’s wrong, what it costs, and a prioritized set of fixes ranked by impact and effort. The result is a concrete artifact your team can act on — not a meeting and a vague verdict.
How engagements run
Engagements are deliverable-shaped and run remote and async — scoped, executed, and delivered, with synchronous time kept to a kickoff and a readout rather than a calendar of standing meetings. It keeps the focus on the work and the artifact, and it tends to move faster. When a live working session is the right tool for a problem, CloudCraft uses it.
Running GenAI on AWS — or about to?
A short conversation is the best way to see whether CloudCraft is the right fit for what you’re building or what it’s costing you.