AI Engineer

10x Labs; Newcastle City Council; On-site; AUD 150,000 - 210,000; Full time

Job summary

10x Labs in Australia seeks a senior AI engineer to design agentic orchestration layers, integrate LLMs and build production-grade AI systems for regulated environments. You will own decisions on agent vs. deterministic code, and work directly with founders to shape scalable solutions.

You will implement agent harnesses, evaluation, and safety guardrails, while applying classical ML for decision-making and ensuring rigorous testing from simulation to live data.

Qualifications

• Deep Python expertise and a track record of building complex systems from scratch.
• Practical experience building agentic workflows or orchestration layers, including LLM integration and evaluation.
• Judgement about when to push agentic and when to fall back on deterministic code for reliability.
• A rigorous approach to testing non-deterministic systems: evals, simulation, historical replay.
• Clear communication and full ownership without micromanagement.

Responsibilities

• Design and build agentic orchestration layers: agent harnesses, tool interfaces, memory and state management, escalation paths
• Integrate and evaluate LLMs against real operational workloads, tested against historical data and simulation before production
• Build the deterministic guardrails: bounded autonomy, audit logging and code-based fallbacks where agents can't yet be trusted
• Apply classical ML (random forests, gradient-boosted models) for decision-making and scoring where it beats generative approaches
• Turn messy operational processes into agent-runnable workflows, working directly with founders and clients
• Help set the standard for how we build, test and ship agentic systems

Skills

Deep Python expertise

Agentic orchestration experience

LLM integration

Deterministic fallback design

Testing non-deterministic systems

Job description

We wrap our arms around businesses and help them scale. We work with enterprises and fast-growing companies to implement AI strategy, workflow automation, and build custom software. Not as a vendor, but as a genuine partner with skin in the game.

We move at a pace most businesses have never experienced, and we hold ourselves to a standard most agencies can't match. We don't pitch solutions. We sit with our clients, understand where they're stuck, and build the systems that get them unstuck. Quickly and properly.

The work is the kind engineers actually want: agent-first platforms running national field operations at around 90% automation, electric vehicle operating systems for everything from cars to trucks and buses, AI making real operational decisions in regulated industries, and rescues of platforms other teams couldn't deliver. And for the right businesses we go further. We unlock them: we build new products we come up with together, and we bring in capital through equity or debt to fund the growth those initiatives create.

Our founders have built energy flex markets, led transactions of billion-dollar businesses, and built software agencies that consistently deliver engineers in the top 3% globally. That pedigree shapes how we think, how we work, and what we expect from the people who join us.

The role

Most AI engineering jobs are demos and pilots. Ours are production systems: agent workforces running rostering, claiming, payroll and compliance workflows end to end, and orchestration layers making operational decisions around the clock in regulated industries. Not chatbots bolted onto a UI. Autonomous systems with audit logs, bounded autonomy and escalation paths, because that's what the work demands.

You’ll build agentic orchestration from the ground up: the harness, the tooling, the evaluation, and the deterministic fallbacks that make agents safe to run in production. You’ll decide when a process should be handled by an agent and when it should fall back to code, and you’ll own that decision and its consequences.

What you will do

• Design and build agentic orchestration layers: agent harnesses, tool interfaces, memory and state management, escalation paths
• Integrate and evaluate LLMs against real operational workloads, tested against historical data and simulation before production
• Build the deterministic guardrails: bounded autonomy, audit logging and code-based fallbacks where agents can't yet be trusted
• Apply classical ML (random forests, gradient-boosted models) for decision-making and scoring where it beats generative approaches
• Turn messy operational processes into agent-runnable workflows, working directly with founders and clients
• Help set the standard for how we build, test and ship agentic systems

What you bring — Required

• Deep Python expertise and a track record of building complex systems from scratch, not configuring platforms
• Practical experience building agentic workflows or orchestration layers, including LLM integration and evaluation
• Judgement about when to push agentic and when to fall back on deterministic code for reliability
• A rigorous approach to testing non-deterministic systems: evals, simulation, historical replay
• Clear communication and full ownership without micromanagement

Strongly preferred

• Classical ML in production decision systems
• Deploying AI into regulated or compliance-heavy environments
• Deep API integration experience: ERPs, payroll, billing, market systems
• Flex market and energy systems experience is always a big bonus

Nice to have

• Contributions to open-source agent frameworks or eval tooling
• Exposure to field-service or logistics operations

Why this role

The systems you build here will be running real businesses within weeks of you joining, alongside people who've genuinely done it before. If you're tired of AI theatre and want your work measured by operations it runs, this is the role.

What you get

• A competitive package that reflects the calibre we hire at
• Access to every AI tool on the market (as long as it has passed security) with team support and budget to get you set up and moving fast
• Hard, interesting problems: real AI in production, real scale, real consequences. Not demos and decks
• Direct access to some of the most experienced operators in the country
• A team that will stretch you, back you, and make you better

The interview process

Our process is straightforward and moves fast: a short video interview (we know engineers hate this — the only people who don't hate it are in sales — but it's an important and necessary step, and if you can't push outside your comfort zone, you're not going to fit in here), then a technical deep-dive on real problems, not puzzles, then a final conversation with the founders. We move quickly for the right person.