Company Description

AIONIX is the intelligence layer for the AI-powered workplace. We connect to all the AI tools organisations use, then turn that data into proactive optimisation: helping organisations improve their AI performance, cost and risk at scale, while giving every employee personalised recommendations to work smarter and get more from their AI tools.

We’re a small team moving at breakneck speed. We don't just need someone to write code; we need an experienced engineer to help us define what "Production AI" looks like.

Fair warning: we move fast and it isn't always tidy. You'll hit half-finished abstractions, decisions made in a Slack thread, and changes that can come up while the code is being written. If that sounds like a reason to leave, this is the wrong role. If it sounds like a reason to join, keep reading.

What you’ll actually do

- Ship, don't just build: You'll own the full AI dev loop - from messy experimentation to production-grade agents that handle real enterprise scale.
- Own how we prove our agents work: Our eval and observability approach isn't decided yet. Build versus buy, what we instrument, what gates a release. That call is yours to make and yours to defend.
- Product-led Engineering: You’ll get deep into customer problems. You aren't just building features; you're solving why a CISO is terrified of "Shadow AI" or why a CFO can't see their AI spend.
- Work Directly With the Founder/CTO and CPO: You'll work day-to-day with the CPO on what we build and why, and report to the CTO. No layers, no proxy.

Who You Are:
- You’re a product-minded builder: You don’t just care if the code works; you care if the customer actually finds it useful.



You’re obsessed with how people interact with AI and will pivot your technical approach to solve a real human friction point.
- You have views and are not afraid to share them: You have solid views on both the tech stack and the product roadmap, you make the case well, and you expect to be heard. We trust the people we hire and we expect you to disagree with us when you think we're wrong.
- You’re a team player: You naturally make the people around you better. You see feedback - both giving and receiving it - as a gift and the fastest way to level up.
- You own the outcome: You don't say "that's not my job." Whether it's debugging a production issue or jumping into a customer discovery call, you take full ownership. You care and are passionate about what you build, and you get a kick out of shipping something that solves a problem people had written off as impossible.

What we’re looking for:
- 6+ years of engineering, with at least 2 spent shipping AI systems to production . If you've got less total experience but unusual depth in production AI, make your case. (Career breaks don't count against you. Tell us about the work, not the timeline).
- You've built agents, not just LLM API calls. Multi-step, RAG, tool-calling, running unattended against real enterprise data.
- Evals are how you work , not a phase at the end.



You build test sets from real failures, judge agent behaviour and not just final answers, calibrate LLM judges before trusting them, and run regressions in CI. "It looks okay" is not a testing strategy.
- You instrument what you ship. Traces across agent and tool calls, cost per run, failure modes you can actually query later. We sell AI cost visibility, so we should be the best in the world at measuring our own.
- Ambiguity is your friend: If you need a 50-page spec to start, this isn't the place. If you like building the spec while you build the code, let's talk.

Current stack (still applies even if you don’t have all of this)

- Serverless architectures: AWS Lambda, API Gateway, SQS/EventBridge.
- Python and TypeScript/Node.js.
- PostgreSQL at scale, including vector search.
- LLM API ecosystem: prompt engineering, token management, streaming, function calling and RAG.
- LangChain/LangGraph/LangSmith ecosystem.
- Production AWS services: ECS, S3, and Bedrock.

Why join us
- We're early, but not speculative. Real enterprise customers already, SOC 2 Type II and ISO 27001 done. This isn't a deck and a dream.
- No legacy to work around. You're not maintaining someone else's decisions. You're writing the foundation the rest of the team will build on.
- The problem is timely. Every enterprise is buying AI faster than it can govern it, and almost none of them can answer what's running, who's using it, or what it costs. We're building the answer.
- Remote-first in Australia

We’re early, we’re ambitious, and we’re building something that matters. Come build AIONIX with us.

📌 AI Software Engineer (Sydney)
🏢 AIONIX
📍 Sydney