Haideyi Technology designs, builds and maintains custom software for companies that need a dependable engineering partner. AI tools are part of our everyday workflow — for coding, testing and review — but everything that ships is tested, reviewed and owned by a human engineer.
We are a compact software company, and we present ourselves that way — because for most projects, a small senior-led team delivers better than a large anonymous one. No account-manager relay, no hand-offs between departments: the people you meet on the first call are the people who build your system.
How we are organised, in plain terms: Haideyi Technology is the service brand of our group. Project delivery is handled by our engineering team in Chengdu, China, where most of our clients and project history are. Our affiliated company in Singapore (incorporated 2025) supports regional business and contracting. Contracting entity details are provided in writing before any engagement — we would rather explain our structure up front than dress it up. Our Chengdu operating company, Chengdu Haideyi Technology Co., Ltd., was founded in 2023 and is publicly registered in Chengdu, China.
Every project is led by an engineer who has shipped production systems before — and who is accountable for the result, not just the timeline.
Documented code, automated tests and infrastructure as code, from day one. Your system stays maintainable — by us, or by your own team later. No lock-in by design.
We will tell you when a project should be smaller, later, or built with an off-the-shelf tool instead. A smaller project that succeeds is worth more than a big one that stalls.
We don't quote AI percentages — they are easy to print and impossible to verify. What we can describe is exactly where AI sits in our pipeline, and where it is not allowed to decide alone.
AI assistants draft boilerplate, scaffolding, migrations and repetitive code. Engineers treat that output as a draft: it is read, corrected and integrated by the person responsible for the feature.
Unit and integration tests are drafted with AI assistance and then extended by engineers for the edge cases that matter to your business. Tests run in CI on every commit.
Every pull request gets both an automated pass (static analysis, dependency and vulnerability scanning) and a human review. Security findings are fixed by engineers, not auto-merged.
API documentation, changelogs and runbooks are drafted from the code and reviewed like code. The result: systems your team can understand and operate after we hand them over.
Four ways of working with us. Each starts with a written scope and a named engineer in charge.
Bespoke systems built around your business process — from first prototype to production and beyond.
A growing part of our business: the hardware and deployment layer underneath AI — for clients who need compute they control, on their own premises or in their own data centre.
Additional engineering capacity that works inside your process — your tools, your rituals, your roadmap.
Independent advice before you commit budget: where AI genuinely helps, and where it doesn't.
Most of our project history is in China, for clients who keep their vendor relationships confidential. Their names and logos are not ours to publish, so there is no logo wall on this site. The descriptions below are real project types from our delivery history; fuller references can be shared during evaluation, subject to client permission.
For a regional retail business: storefront, ordering and payment, inventory and logistics integration — built to their process rather than adapted from a generic platform, and iterated with their team after launch.
For a logistics operator: shipment and warehouse workflows, tracking, internal settlement and reporting — replacing spreadsheet-based coordination across sites.
For a large enterprise group: approval workflows, document flow and management reporting in one internal platform, rolled out department by department.
Planning, equipment deployment and operations support for enterprise data-centre facilities, including monitoring setup and documented handover to the client's own operations staff.
Deployment and ongoing operation of GPU compute for AI training and inference — capacity planning, cluster setup and the unglamorous operational work that keeps expensive hardware actually utilised.
For local enterprise clients: systems planning, architecture review and delivery support for custom development programmes.
A plain process you can follow from outside. You always know what is being built, what it costs, and what is next.
We map the problem, the users and the constraints — and agree on what success looks like.
A written proposal: scope, architecture, timeline and price. No surprises later.
Short sprints with a working demo every week. Feedback changes the plan while it is still cheap to change.
Automated tests, human review and acceptance testing with your team on a staging system.
Deployment, monitoring and documentation — then ongoing support if you want it.
Our delivery history is concentrated in a few areas — most of it with clients who value documented delivery, formal acceptance processes and long-term support.
Custom systems built to specification, with the documentation and acceptance process these clients require.
Operational systems for moving and tracking goods.
Storefronts and the operations behind them.
A short list, deliberately. These are the tools we know deeply and can support for years — not a catalogue of everything we have ever touched.
A few sentences are enough: what you want to build, when you need it, and what you have tried so far. You will hear back from an engineer, not a sales script.
Haideyi Technology · Engineering in Chengdu, China · Singapore affiliate for regional contracting