Artificial Intelligence Engineer

Property Dynamix; Dubai; Hybrid; AED 300,000 - 520,000; Full time

Job summary

Property Dynamix in Dubai is hiring an AI Engineer to design multi-agent systems that turn market data into actionable insights for real estate pricing and research. You will build tool-using agents, ensure accuracy with grounding checks, and optimise production deployments on Google Cloud.

You’ll work on data-heavy tasks, from BigQuery SQL to tool schemas, while reducing latency and cost, with a hybrid remote-friendly setup.

Qualifications

• Experience building production-grade LLM applications with tool calls and agents.
• Strong Python, TypeScript and SQL skills for large datasets.
• Ability to design evaluation metrics and scorecards for model outputs.
• Experience deploying and debugging on a cloud platform (prefer Google Cloud).
• Ability to explain failures and justify fixes with data.

Responsibilities

• Design and build multi-agent workflows using Google's Agent Development Kit (Python).
• Implement tool calling with Gemini and define tools and schemas.
• Write and optimise BigQuery SQL over millions of transactions.
• Create guardrails and grounding checks to prevent invented outputs.
• Develop evaluation harnesses and golden test sets to measure MOIRA improvements.
• Manage model costs, latency, and production deployment on Google Cloud.
• Extend Majlis connector for Claude and ChatGPT.

Skills

Python

TypeScript

LLM apps

SQL/BigQuery

Evaluation/QA

Cloud deployment

Troubleshooting

Education

CS degree or equivalent

Tools

Google Cloud

Firebase

LangGraph

Job description

Priority will be given to Candidates - Experienced in Real Estate Pricing, Real Estate Research UAE.

Type - Hybrid/Remote

Company Description

Property Dynamix is a real estate operating system for developers in Dubai and Abu Dhabi. We turn registered DLD and ADREC transactions into defensible decisions: unit pricing, pricing simulations, payment plans, land feasibility, new launch tracking, promotion monitoring and monthly market reports.

Most platforms in this market sell data. We build what happens after it. A team of AI agents researches comparables, checks pricing against the market, and reviews its own work before a user sees it, with every figure traced back to registry records.

Our AI analyst, MOIRA, answers market questions in plain language. Through Majlis, our connector, the same market data is available inside Claude and ChatGPT.

Role Description

This is a full-time, Hybrid/Remote AI Engineer role in Dubai.

You won't be training neural networks from scratch. You'll build the agent systems that turn market data into answers people can act on, and the checks that keep those answers honest.

What you'll work on:

• Designing and building multi-agent workflows on Google's Agent Development Kit (Python), where each agent has one job and the results are orchestrated into a single answer
• Tool calling with Gemini: defining tools, handling multi-round tool loops, structured outputs and response schemas
• Writing and optimising BigQuery SQL over millions of registered transactions, and turning them into the tools agents call
• Building guardrails that stop an AI from inventing numbers: grounding checks, claim verification, and gates that reject output a person should never see
• Building evaluation harnesses and golden test sets, so we measure whether a change made MOIRA better or worse instead of guessing
• Managing model cost and latency: choosing the right model per task, routing calls, and metering spend
• Shipping to production on Google Cloud: Cloud Run jobs, Firebase Cloud Functions, scheduled pipelines
• Extending Majlis, our MCP connector for Claude and ChatGPT

Qualifications

Must have:

• Strong Python, and working TypeScript
• Hands-on experience building LLM applications in production: tool calling, agents, structured output, prompt design
• Solid SQL, ideally BigQuery, on large datasets
• Experience with evaluation: you've built tests or scorecards for LLM output and used them to make decisions
• Comfortable deploying and debugging on a cloud platform, ideally Google Cloud
• You check your own work. You measure before you fix, and you can explain why something broke, not just that it now works

Nice to have:

• Google ADK, LangGraph or a similar agent framework
• Firebase (Firestore, Auth, Cloud Functions)
• Experience with real estate, pricing or financial data
• Knowledge of the Dubai or Abu Dhabi property market

• Degree in Computer Science, AI, Data Science or a related field, or equivalent experience shipping real systems.