This is a hands-on engineering role focused on building the product layer of AI features, such as agents, retrieval and tool use, on top of an existing grid simulation platform. A solid candidate is a solid backend or full-stack engineer with at least five years of experience who has shipped AI features to paying customers and kept them running after launch. They also bring real evaluation habits and good judgement about where AI should not be trusted.

How it fits the AI era

AI lets far more people get answers from a deep engineering platform, in days instead of months of specialist work. The work here is making those answers grounded and traceable, since a confident wrong answer about pole loading or clearance is a safety issue. What stays human is the engineer who signs off, the judgement on what the system may do, and the conversations with expert users who spot errors fast.

How it builds AI experience

The role offers direct experience building agents and retrieval systems grounded in a physics-based digital twin,



which is a demanding setting for testing AI reliability. It also covers evaluation sets, guardrails, refusals and human review steps for critical infrastructure. Someone short on this could build a small retrieval or agent project over public engineering or geospatial data and write up how they tested it and where it failed.

Worth showing

- Shipped LLM features to real users
- Evaluation and regression testing approach
- Retrieval and agent orchestration
- Guardrails for high-stakes output
- Working with expert users

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About Neara

A physics-based 3D digital twin of the electricity network, so utilities can simulate a storm or a fire before it happens rather than inspect poles after.

#J-18808-Ljbffr

📌 AI Product Engineer (Sydney)
🏢 present of AI
📍 Sydney