Data platforms · ML systems · Custom software

Engineering for the
data you already have

We build the warehouses, pipelines, models and applications companies need once spreadsheets and off-the-shelf tools stop holding. You keep the repository, the architecture decisions and the running system.

Start a projectSee how we work →

New Zealand + remoteFixed-scope or embeddedNDA-friendly

siyon.co.nz / pipeline
An illustration of a data pipeline. Sources feed an ingest stage, then transform, model and serve stages, with each stage lighting in turn as a batch moves through. Alongside it: incremental loads, tested models and freshness alerts.

What we do

Four services, one engineering team

The same people design the platform, train the models, build the application and answer the pager. Nothing is handed to a second vendor halfway through.

Move your data once, correctly, and stop rebuilding it.

We design the warehouse, write the pipelines that fill it, and leave behind tests that tell you when a number is wrong before your board does.

What you get

  • Warehouse and lakehouse builds on Postgres, BigQuery or Snowflake
  • Orchestrated pipelines in Airflow, dbt and Kafka
  • Data quality tests and lineage you can audit

You usually call us when

  • Reporting runs off exports that someone refreshes by hand
  • Two systems disagree and nobody can say which is right
  • A pipeline exists but breaks silently and nobody owns it
Explore Data Engineering →

Postgres · BigQuery · Snowflake · dbt · Airflow

Models that reach production, with the monitoring to keep them honest.

Most models die in a notebook. We build the ones that ship, served behind an API, measured against a baseline, and watched for the day the data shifts underneath them.

What you get

  • Forecasting and demand models
  • Document and LLM pipelines with retrieval and evals
  • Deployed inference with drift monitoring

You usually call us when

  • A notebook proved something works, but nothing runs on a schedule
  • Forecasting happens in a spreadsheet one person maintains
  • An LLM demo impressed everyone and nobody can measure whether it is right
Explore Data Science & ML →

Python · PyTorch · scikit-learn · XGBoost · MLflow

Web applications and APIs built to be handed over.

We build the internal tools, customer portals and integrations that off-the-shelf software will not cover. Work happens in your repository, on your cloud, documented well enough that another team could take it tomorrow.

What you get

  • Python/FastAPI and TypeScript/React builds
  • API and integration work
  • Cloud deployment, CI/CD, and observability

You usually call us when

  • A critical process runs on a spreadsheet and one person's memory
  • Two systems need to talk and the vendor quote for the connector is absurd
  • An internal tool exists but nobody remembers how to deploy it
Explore Software Engineering →

Python · FastAPI · TypeScript · React · PostgreSQL

The team that built it also keeps it up.

Software does not stop needing attention at launch. We take the pager for systems we built and, after a review, for systems we did not.

What you get

  • SLA-backed application support
  • On-call and incident response
  • Platform upgrades and cost tuning

You usually call us when

  • The person who built it has left and nobody else can deploy
  • Something fails overnight and the first alert is a customer email
  • Cloud spend keeps rising and nobody can point at what changed
Explore Managed Support →

PagerDuty · Grafana · Prometheus · Sentry · Datadog

Rockit Desk · Built and run by SiyonIT

A full service desk.
Without the rollout project.

Rockit Desk is our service-management platform: incidents on a live SLA clock, problems, changes with approvals, a CMDB, a knowledge base and a self-service portal. It is ITIL-aligned where your auditors need it to be, and you configure it from admin screens rather than filing a change request with a vendor.

See Rockit Desk →

SiyonIT builds it, runs it and supports it.

  • A board that watches the clock

    Every open ticket, its owner and its SLA timer on one live board, with routing, escalations and breach warnings that fire on their own.

  • The whole practice set

    Incidents, problems, changes with approvals, service catalogue, request fulfilment, knowledge and CMDB, with audit history on every record.

  • Yours to change

    SLA targets, catalogue items, approval rules and permissions sit in admin screens you control, with a REST API, webhooks and SSO for everything else.

  • Not only IT

    HR, Finance, Facilities, Legal and Procurement can each run a service workspace on the same board and the same SLA engine.

How we work

Four stages, and what you receive at each

  1. Discover

    One week, fixed fee. Ends in a written technical plan you own, even if you stop there.

  2. Design

    Architecture, data contracts and estimates, reviewed with your engineers before any build starts.

  3. Build

    Two-week increments with a demo at the end of each one. Your repository from day one.

  4. Run

    Handover documentation and a walkthrough with your team, or we stay on for support.

Stack we use

Standard tools, chosen so you can hire for them

Nothing exotic. Everything here has documentation and a hiring market, so the platform outlives the engagement.

Data

Warehouses, pipelines, streaming
  • Postgres
  • BigQuery
  • Snowflake
  • dbt
  • Airflow
  • Kafka
  • Spark

ML

Training, serving, evaluation
  • Python
  • PyTorch
  • scikit-learn
  • MLflow
  • LangChain

Platform

Applications, deployment, runtime
  • FastAPI
  • React
  • TypeScript
  • Docker
  • Kubernetes
  • AWS
  • GCP
  • Cloudflare

Engagement models

Three ways to work with us

Questions we get asked first

Who owns the code and the IP?

You do, from the first commit. Work happens in your repository under your organisation, and the contract assigns all IP to you. We keep no licence over what we build for you.

Do you work with our existing team?

Usually, yes. We join your standups, review your pull requests and follow your conventions. Where your team wants to own a component, we hand it over with documentation rather than keeping it.

Can you take over a half-built system?

Often. We start with a paid review of the codebase and infrastructure, then tell you plainly what is worth keeping and what is cheaper to rebuild. If the honest answer is that you do not need us, we will say so.

How do you handle data privacy and NDAs?

We sign your NDA before the first technical conversation. Wherever possible we work against anonymised or synthetic data, in your cloud accounts, with access scoped to named engineers and revoked at the end of the engagement.

What if we only need advice?

Then buy the Discover week on its own. You get the written technical plan, the architecture options and the cost estimates, and you are free to build it with your own team or another vendor.

What timezones do you cover?

We work New Zealand hours and overlap with Australia and South-East Asia the same day. For clients further west we agree a fixed overlap window in writing rather than promising round-the-clock cover we cannot staff.

Tell us what is breaking

Describe the system, the data and the deadline. We will reply with an honest read on scope and whether we are the right team for it.

Start a project

or email hello@siyon.co.nz