Skip to content

01 / Services

What we are actually good at.

Five practices. They overlap on purpose — most of the problems worth hiring for sit between two of them.

Practices

The detail.

No packages, no tiers. Engagements are scoped against the problem.

01

Product Engineering

Production web applications, from the database schema to the deployed frontend.

  • Typed end to end. React and TypeScript on the client, NestJS and Prisma against PostgreSQL on the server.
  • Multi-environment by default: isolated dev, staging and production, promoted through CI rather than by hand.
  • We take ownership of systems that already exist as often as we start new ones.
  • TypeScript
  • React
  • NestJS
  • Prisma
  • PostgreSQL
  • Python
02

Cloud & Infrastructure

AWS architecture, deployment pipelines, and the unglamorous networking underneath.

  • Application hosting, object storage, managed databases, WebSocket APIs, and the IAM boundaries between them.
  • GitHub Actions pipelines that promote a branch to staging and staging to production, with encrypted configuration at rest.
  • Private networking, SFTP endpoints for partner data exchange, DNS and mail authentication done correctly.
  • AWS
  • GCP
  • GitHub Actions
  • Terraform
  • SFTP
  • WireGuard
03

AI Integration

Language models wired into real workflows, with measured accuracy instead of demos.

  • Document extraction pipelines that turn PDFs into validated, schema-checked records a human confirms in seconds.
  • Every pipeline ships with an evaluation harness: ground truth sampled from production, per-field accuracy, and cost per document.
  • If a model cannot beat the manual process on both accuracy and cost, we say so rather than ship it.
  • Claude API
  • Zod schemas
  • Eval harnesses
  • RAG
  • OCR
04

Data Platform

Pipelines, orchestration and the reporting layer the business actually trusts.

  • Ingestion and transformation pipelines running on managed cloud services, deployed per application rather than as one monolith.
  • Webhook and partner integrations handled as first-class services, with replay and failure visibility.
  • Warehouse modelling and BI, so that two teams asking the same question get the same number.
  • Python
  • FastAPI
  • GCP
  • BigQuery
  • Cloud Run
  • Superset
05

FinOps

AWS spend brought back under control, then kept there.

  • Full cost attribution first. You cannot cut what you cannot attribute to a team, an environment or a feature.
  • Rightsizing, storage lifecycle, commitment coverage and the idle non-production resources nobody remembers provisioning.
  • Budgets, anomaly alerts and a reporting cadence, so the savings survive the quarter they were found in.
  • Cost attribution
  • Rightsizing
  • Savings Plans
  • Anomaly detection
  • Tagging