About the Role

Build the AI layer of a UAE-based, AI-powered wellbeing platform that helps people improve their wellbeing across six dimensions: Physical, Mental, Social, Nutritional, Professional and Financial. Its AI companion combines wearable data, assessments, behavioral signals and personalized recommendations.

This is a data engineering role with real data science depth. You will own the data foundation behind the AI (pipelines, datasets, retrieval and feature layers) and build and ship the models that run on top of it. The role is remote for candidates based in Egypt, and we need someone who can join immediately.

Experience

5+ years in Data Engineering, ML Engineering or Applied AI, including at least 2 years building data for ML or LLM systems in production.

Key Responsibilities

Data platform and pipelines

• Design and build batch and streaming pipelines for wearable, assessment, app engagement and user feedback data
• Model data in PostgreSQL and the analytics layer so it is ready for ML, LLM and reporting use
• Build feature pipelines and a reusable feature layer for personalization and predictive models
• Own data quality, validation, lineage and pipeline monitoring

LLM data and fine-tuning

• Collect, clean, label and version English and Arabic datasets for supervised fine-tuning
• Support fine-tuning of open-weight LLMs (Qwen, Llama, Jais and similar) using LoRA, PEFT and SFT
• Build evaluation datasets and benchmarks for model selection in English and Arabic

RAG, agents and contextual memory

• Build ingestion, chunking and embedding pipelines for the AI companion's knowledge base
• Run and tune vector storage and retrieval for accuracy and latency
• Build the data layer for tool calling and per-user contextual memory, connected to platform services and user data

Data science, personalization and recommendations

• Develop recommendation engines from wearable data, assessments and user behavior
• Build predictive models on time-series and behavioral data to surface trends and wellbeing improvements
• Run experiments and A/B tests, and feed engagement and feedback data back into model improvements

Production deployment and MLOps

• Work with Engineering and DevOps to deploy data and AI services on UAE-based infrastructure (Azure UAE North)
• Contribute to hybrid AI architecture, model routing and inference optimization
• Build CI/CD for pipelines and models, and monitor quality, drift, latency and resource cost

Data privacy and AI quality

• Apply anonymization, pseudonymization and access controls to sensitive health-related data
• Keep data handling compliant with UAE PDPL, KSA PDPL and client security requirements
• Design evaluation frameworks for response quality, factual accuracy and personalization

Required Skills

Python, SQL, Airflow / Dagster / Prefect, Spark / pandas / Polars, Kafka or equivalent streaming, PostgreSQL, data warehouse or lakehouse modeling, PyTorch, Hugging Face Transformers and Datasets, fine-tuning datasets (LoRA, PEFT, SFT), RAG, embeddings, vector databases (pgvector, Qdrant or Azure AI Search), statistics, recommendation systems, predictive modeling, cloud infrastructure, GPUs and inference optimization

Requirements

Must have (applications that do not meet these will not be considered)

• Native or fully fluent Arabic, spoken and written. You will build and review Arabic datasets and evaluate Arabic model output
• Based in Egypt and eligible to work there full-time
• Available to join immediately
• Fluent professional English, spoken and written

Also required

• A track record of production-ready data and AI systems, not only research prototypes
• Strong problem-solving and the ability to work independently with a remote team

Nice to Have

• Experience in healthtech, digital wellbeing or wearable technology
• Time-series and behavioral analytics experience
• Arabic NLP, including Gulf and Egyptian dialects, and multilingual language models
• Azure AI Foundry, Azure ML, AWS or self-hosted AI infrastructure
• vLLM, Docker, Kubernetes, MLflow or dbt
• AI safety, model evaluation and monitoring

Expected Outcomes in the First 6 Months

1. Production pipelines for wearable, assessment and engagement data, with quality checks and monitoring
2. English and Arabic datasets plus an evaluation harness used to select and fine-tune models
3. The RAG ingestion pipeline and contextual memory data layer, live in production
4. First personalized insights and recommendation models shipped to users
5. Data and AI services deployed and monitored with Engineering on UAE infrastructure
6. Agreed benchmarks for model quality, latency and cost