Skip to content

The work, without the logos.

Most of our engagements sit inside a customer’s product or infrastructure and are covered by an NDA. We walk through detailed case studies — architecture, decisions, outcomes — on a call. Below is the shape of the work.

01

AI feature to production

AI engineering
The problem
A retrieval prototype worked in the demo and fell over on real documents and real load.
What we changed
An evaluation set wired into CI, tracing and per-request cost reporting, a runbook for drift and rollback, and the data pipeline underneath made ready.
Stack
Python · pgvector · OpenTelemetry · GitHub Actions
02

Enterprise deployment

Enterprise deployment
The problem
A SaaS product that only ran on the vendor's own cloud, blocking an enterprise deal that required a customer-hosted install.
What we changed
One packaged release for AWS, Azure, GCP and bare metal, an air-gapped install path, and the security evidence a procurement review asks for, answered up front.
Stack
Helm · Terraform · OCI images · private registry
03

Zero-to-one build

Product engineering
The problem
An idea with engineering capacity behind it but no decision on what the first version was or how it reached production.
What we changed
Discovery, architecture and a build in two-week cycles, handed over as a production-grade system with runbooks and an infrastructure repository the team owns.
Stack
TypeScript · Postgres · Terraform · CI/CD
The problem
A cloud bill growing faster than usage, with no allocation and no owner against any line of it.
What we changed
Spend made legible per team and per service, waste removed from the architecture rather than a spreadsheet, and budgets and anomaly alerts wired into CI.
Stack
Kubernetes · OpenCost · savings plans · budgets API

Want the detail?

Book a call and we’ll walk through a relevant engagement — the architecture, the trade-offs, and what came out the other side.

Book a technical call