Platform Engineer

YAS; Hong Kong; On-site; HKD 600,000 - 900,000; Full time

Job summary

AURA by YAS is seeking a Platform Engineer to own the end-to-end platform for our risk intelligence system. The role combines data pipeline, infrastructure, and product-facing interfaces, with a focus on production reliability and practical delivery for insurers and fleet operators.

You will work in a small, execution-focused team in Hong Kong, contributing to live operations, automation, and AI-assisted engineering practices. A strong sense of ownership is essential.

Responsibilities

• Own the platform end to end across data pipeline, infrastructure and product layers.
• Operate a production platform used by insurers, fleets and partners.
• Collaborate with AI agents and developers to ship practical, automated solutions.

Job description

We’re looking for an exceptionalPlatform Engineerto help build AURA, our risk intelligence platform for physical AI. This is a full time role requiring Hong Kong work location eligibility.

If someone comes to mind, I’d be grateful for an introduction. And if this sounds like you, I’d love to hear from you. Hong Kong residence and/or working visa is required.

AURA by YAS

AURA makes it possible to price risks that traditional insurance systems were never designed to understand.

We turn live machine data from EV fleets, robotaxis, autonomous systems and robots into risk intelligence that insurers can actually underwrite.

This is becoming one of YAS’s core technology assets. It already supports business in production today and will power the simulation, pricing and risk models we are building next.

We are looking for someone who can own this platform end to end.

The role is roughly70% platform engineering and 30% product engineering, working within a small, highly execution focused team without layers of product management between you, the customer and the problem.

What you’ll own

1. The data pipeline

Build and operate the ingestion, normalisation and contextualisation layer that turns heterogeneous machine telemetry into reliable, scored and context aware events.

Data may come from dashcams, EV fleets, vehicle APIs, robot consoles and other physical AI systems.

You will own the connectors, schemas, data quality and correctness of the underlying record.

2. The infrastructure

Run a production platform that insurers, fleet operators and partners depend on.

That includes infrastructure as code, CI/CD, observability, security, data residency, event infrastructure, geospatial data stores and production reliability.

When something breaks, we need to know why.

More importantly, we want to design the system so the same problem does not need to be solved twice.

3. The product layer

Build and maintain the interfaces used by insurers and fleet operators.

You will work across the AURA API, underwriting and risk query endpoints, AI agent interfaces, internal tools and selected user facing applications.

You are not simply maintaining infrastructure behind someone else’s product. You are helping shape the product itself.

4. Live operations

This is a real production business, so part of the role involves dealing with reality.

A new fleet may require a bespoke integration.

A vendor may change an API without warning.

Two datasets may stop reconciling.

An insurer may need information urgently.

The board or an investor conducting due diligence may need a dataset pulled and validated.

Dependencies need patching. Infrastructure needs upgrading. Systems occasionally behave in ways nobody expected.

We would rather tell you this now than have you discover it four months later.

The expectation is simple: solve the immediate problem, then automate what is likely to happen again.

How we work

You already work with coding agents.

For a team our size to cover this much surface area, AI assisted engineering is part of how we operate.

We care less about which tools you use and much more about your judgement.

What do you delegate?

What do you write yourself?

How do you review agent generated code before it touches pricing, premium, claims or customer data?

You will also help structure the codebase so agents can work effectively within it through strong tests, clear architecture, good documentation and fast feedback loops.

And because AURA increasingly exposes intelligence directly to AI agents, you will be buildingfor agents as well as with them.

Nobody will hand you a perfect specification.

Requirements may start with a conversation with an underwriter, fleet operator, founder or partner.

You will often be in that conversation.

You need to understand the problem, challenge assumptions and translate it into something that works.

If you need a fully groomed backlog before you can start, this probably will not be the right environment.

Glue before build.

Where possible, buy it, integrate it and get it running.

Build custom infrastructure when it creates genuine differentiation.

Ship something practical first. Improve it once the value is proven.

We do not have the headcount or interest to engineer beautiful systems nobody uses.

You will learn insurance.

We do not expect you to arrive knowing all of it.

We do expect you to become curious enough to understand why the system you are building matters commercially.

Direction will change.

We are building in a market that is still being created.

AURA is evolving from a pricing and risk engine into infrastructure that insurers, fleets and machines can query directly.

Something built in Q1 may take a different form by Q3.

You will not simply execute against that change. You will help decide what changes.

You are measured by what runs.

Not what was proposed.

Not what was designed.

Not what is 80% migrated.

In a small team, working software matters.

If you are the kind of engineer who enjoys ownership, ambiguity, production responsibility and building things at the intersection ofAI, physical machines, data and insurance, we should talk.