We build and operate open-source inference.
Northwood is the deployment partner for open-source inference. We build the open-source tooling that routes, prices, and monitors production AI, then deploy and operate it in your environment: the best-fit model for each request, lower cost, and every decision on record.
We work with AI-native companies, investment teams, and technology companies that run real inference in production and need it to cost less, hold its quality, and stay accountable.
Northwood is the accountable partner for the full inference stack: providers and compute, models and routing policy, the open-source control layer, implementation, and ongoing operations. We’re open-first and model-agnostic: the best open models by default, commercial where it earns its place. The goal is production AI that costs less and answers for itself, not a single product, for teams where cost, quality, and accountability are non-negotiable.
Built from enterprise AI deployment experience.
Northwood Systems was founded by Andrew Liu, a Toronto-based AI engineer, technology leader, and former co-founder and CTO of Centro.
Before Northwood Systems, Andrew helped bring Centro's AI technology into enterprise operations across supply chain, retail, and advanced manufacturing, working with teams that needed AI to fit their workflows, their data, and their operating constraints.
Centro was acquired by Rose Rocket in 2025, and Andrew joined Rose Rocket as a Staff AI Engineer to help advance AI adoption inside transportation and logistics software.
Northwood Systems builds on that experience: deploying practical AI systems for complex organizations that need to move faster while maintaining control, trust, and operational discipline.
We start with your inference, then deploy the control layer around it.
Every engagement begins by mapping your current inference spend, workloads, quality requirements, and constraints. From there, we deploy and operate the open-source control layer around your actual production traffic.
Map before you build.
We document your inference spend, workloads, quality bars, and constraints before provisioning a single server. You get a written deployment brief you own.
Your environment, your rules.
Every Northwood Systems deployment operates inside your infrastructure, your identity provider, and your change-management process. Nothing reaches an external service unless you explicitly approve it.
Production-ready in one month.
By the end of week four, your team has a running control layer, a staged production route, trained operators, and a playbook for expanding to new workloads.
Your knowledge, your models, your AI deployment.
Northwood helps technical teams move from a single model or prototype to governed, model-portable inference. Our open-source projects provide the control layer; Northwood deploys and operates the complete system.