Tired of "the demo worked" being the finish line?

As an AI Engineer at Liberate, it's about so much more than getting a model to say something clever. It's a given that you can wire up a foundation model, chain a few tool calls, and make a demo sing. The question we actually care about, the one most places wave past, is whether the thing works on the thousandth try, not the third. And whether you can prove it. Most environments let prompt engineering be the finish line. A response that looks about right, a shrug, ship it. If you've watched "it ran fine when I tried it" get treated as done and known it wasn't, you'll fit right in here. We treat evaluation as the heartbeat of the product, not a box to tick before launch. The goal isn't a better benchmark score. The goal is a better product. We care about the difference between a change that made things genuinely better and one that just made them different, because those two are far easier to confuse than most people realize.

Who we are - A real, estate story.

This is a story about unlocking potential, together. The UAE real estate market has become awash with agents lacking knowledge and etiquette. Why? Everyone and their uncle is trying their hand at real estate, hoping to make a quick buck. The barriers to entry are low, and traditional brokerages, in their race to expand, have opened the floodgates of the industry to mistrust and misinformation. We champion the best 1% of agents. Their expertise, their dedication to quality service, and their long-term commitment to their market. These are the industry leaders. We support these gallant warriors to create and grow their own companies. Companies that put them, their knowledge and values at the forefront. Companies that are supported by services that are built not for vanity, but to optimize the agent and client experience. Our vision is a network, a guild of experts, collaborating with undoubted trust and without duplicity - which we believe will revolutionize the way property is transacted. Liberate's mission is to create a more sustainable ecosystem for this wondrous and often misunderstood asset class to prosper.

Your responsibilities?

As an AI Engineer at Liberate, you will own whether our AI stack actually works, not just whether it runs. You will work directly with the founding team, product, engineering and our agents to understand what the product needs to get right, where it currently falls short, and how we would even know. You will be expected to own the full loop: how we measure, how we label, how we build, and how the system improves over time. We do not believe in clean hand-offs or siloed responsibilities. You will likely find yourself in situations where the problem is ambiguous, the data is incomplete, the ground truth is uncertain, and the answer is not obvious. That's OK.

What we expect from you

We are looking for someone who has seen both sides: the applied ML discipline of evaluation, data and measurement, and the practical reality of shipping LLM-powered systems that real people depend on. You will work on Libbie 🙋‍♀️, our AI enabled “Broker’s Best Friend”. It spans real-time conversation classification and a conversational engine that has to understand a request and return the right result, consistently. Both live or die on the same thing: our ability to prove they work, and to make them better over time. You should have real roots in machine learning. You should care about how a model is measured, how a dataset is built and trusted, and how you know a change made things genuinely better rather than just different. But you should also be practical. You should have shipped things. You should know that a model in a notebook is not a product, and that the gap between the two is where most of the real work lives. This is not a role for someone who stops at a working prompt. It is for someone who wants to build the thing that tells us whether any of it is actually working.

• Strong grounding in applied machine learning, with real depth in evaluation and data-centric methods
• Experience designing evaluation for production systems, including metrics, agreement, significance and consistency across runs
• Experience building and validating labelled datasets, and scaling labelling beyond manual effort
• Hands-on experience with production LLM systems and agentic loops, enough to architect, build and operate end-to-end
• Sound judgment on where to use a large model, a smaller fine-tuned model, or deterministic logic you can test
• Experience taking systems from prototype to production, with real attention to reliability, latency and cost
• Ability to raise the level of strong engineers who are newer to this space, without a formal reporting line
• Ability to reason from first principles rather than defaulting to the popular approach
• Clear written and spoken communication, especially explaining technical and statistical ideas simply
• Fluent written and spoken English
• A level of rigor and curiosity that raises the bar for everyone around you