AI Engineer

Location: Dubai, United Arab EmiratesEmployment: Full-time, on-siteCompany DescriptionMajlis CRM is an AI-powered real estate CRM built for brokerages in the UAE. We help real estate teams manage leads, match buyers with properties, automate follow-ups, and improve client engagement. We are building AI into everyday brokerage workflows, giving agents more time to build relationships and close deals.Role DescriptionWe are looking for an AI Engineer to design, build, and deploy production-ready AI features across Majlis CRM. You will work closely with our founders, product team, and developers to turn real estate workflows into reliable AI-powered experiences.Your work will include conversational assistants, AI agents, intelligent property search, lead qualification, and call and meeting intelligence. You should be comfortable taking a feature from experimentation through deployment, measuring its performance, and improving it based on real customer usage.Key ResponsibilitiesBuild and integrate features using leading language and multimodal models from OpenAI, Anthropic, Google, and relevant open-weight model providers.Develop AI agents that retrieve information, use tools, update CRM records, create tasks, and coordinate follow-ups with appropriate permissions and human approval.Build retrieval-augmented generation (RAG) pipelines, including document ingestion, parsing, chunking, embeddings, hybrid search, metadata filtering, and reranking.Develop natural-language property search and buyer-property matching that combine semantic understanding with accurate structured filters such as budget, location, bedrooms, and availability.Extract structured information from enquiries, conversations, and documents to support lead classification, qualification, routing, and CRM updates.Build call and meeting intelligence workflows covering transcription, speaker diarization, summaries, action items, and follow-up tasks.Design prompts, context management, conversation memory, structured outputs, and tool integrations for reliable AI behavior.Establish evaluation datasets and automated tests to measure retrieval quality, factual accuracy, tool execution, and business outcomes.Monitor production AI systems and optimize response time, token usage, cost, and reliability through model selection, caching, routing, retries, and fallbacks.Implement access controls, tenant isolation, audit trails, and safeguards against prompt injection and unauthorized data access.Evaluate new models and tools, adopting them when they offer measurable improvements for the product.Required QualificationsStrong software engineering fundamentals and proficiency in Python, with experience building backend services, REST APIs, asynchronous workflows, and database integrations.Hands-on experience shipping and maintaining LLM-powered applications beyond prototypes or chatbot demos.Practical experience with one or more leading model APIs, such as OpenAI, Anthropic Claude, or Google Gemini, and the ability to compare models for quality, latency, cost, and privacy requirements.Strong understanding of prompt and context engineering, function calling, structured outputs, schema validation, and model limitations.Experience building agent workflows using direct model SDKs or frameworks such as LangGraph, OpenAI Agents SDK, or equivalent.Experience with RAG, embedding models, vector search, and relational databases, using tools such as PostgreSQL with pgvector, Qdrant, or Pinecone.Ability to decide when to use SQL queries, conventional application logic, retrieval, or an LLM to solve a problem reliably.Experience evaluating and debugging AI systems, including tracing model and tool calls, analyzing failures, and maintaining regression tests.Familiarity with Git, Docker, CI/CD, automated testing, and deployment on a major cloud platform.Strong problem-solving skills, clear communication, and the ability to own features from design through production.What We ValueWe value engineers who can demonstrate working AI products, explain their technical decisions, and show how they measured and improved results. Experience with every listed tool is not required. Strong engineering judgment, practical delivery, and the ability to learn quickly matter most.