Welcome to MultiBank Group, a global financial pioneer established in 2005 in California and now proudly headquartered in Dubai, UAE. We specialize in delivering cutting-edge trading technology, unparalleled liquidity, and exceptional customer service. Our extensive range of financial products includes Forex, Metals, Shares, Indices, Commodities, and Cryptocurrency CFDs.

Join our thriving community of over 2 million clients across 100 countries, contributing to a daily trading volume exceeding US$ 35 billion. As a heavily regulated institution with oversight from 18+ financial regulators across 5 continents, and recipient of over 80 financial awards, MultiBank Group is devoted to innovation, excellence, and empowering our clients to achieve their financial goals.

Role Overview

We are seeking a Senior Data Scientist to join our AI and Machine Learning team. The role sits at the intersection of machine learning, computer vision, and large language models, with a focus on delivering production-grade intelligent solutions within a fintech environment. The successful candidate will contribute to AI strategy, lead end-to-end model development, and work closely with cross-functional teams across data engineering, software engineering, and business functions.

Key Responsibilities

Design, develop, and evaluate data-driven algorithms across classification, detection, segmentation, regression, and anomaly detection, applying both classical and deep learning approaches. Rapidly prototype solutions and evaluate their performance against business objectives

Prototype and assess LLM-based and multimodal systems for document understanding, knowledge extraction, information retrieval, and workflow automation, including fine-tuning foundation models, building RAG pipelines, and extending models for domain-specific applications

Design and implement agentic AI systems including task-oriented agents, workflow orchestrators, tool-using agents, and autonomous reasoning frameworks. Translate complex business workflows into reliable, observable, and maintainable AI-driven pipelines

Own the full machine learning lifecycle from data collection, preparation, and cleaning through model training, evaluation, deployment, and ongoing production maintenance. Champion best practices in MLOps, versioning, and reproducibility

Contribute to solution architecture and collaborate closely with data engineers, software engineers, and domain experts to integrate AI-enabled products into existing systems

Establish robust monitoring frameworks to evaluate AI solution performance post-deployment. Proactively identify data quality issues, model drift, and performance degradation, and drive continuous improvement initiatives

Stay current with advances in AI research, mentor junior data scientists, contribute to internal knowledge sharing, and support the broader AI community of practice within the organization

Requirements

5 to 10 years of hands-on experience in classification, detection, and segmentation using both classical and deep learning approaches, applied to real-world, production-grade problems

Proven track record of developing, deploying, and scaling end-to-end ML pipelines in industrial or enterprise contexts

Hands-on experience building and deploying LLM applications including models such as GPT, Llama, Falcon, and Claude, covering fine-tuning, RAG systems, domain adaptation, and multimodal extensions

Experience designing and implementing agentic AI systems including task-oriented agents, workflow orchestrators, or autonomous reasoning frameworks

Experience collaborating in cross-functional teams and communicating technical outcomes to non-technical stakeholders

Strong foundation in applied mathematics, probability, and statistics underlying modern ML and DL methods

Advanced Python programming skills with a focus on clean, production-ready code

Deep knowledge of ML algorithms and DL architectures including CNNs, Transformers, Diffusion models, and Graph Neural Networks

Proficiency in prompt engineering, evaluation frameworks, and structured output design for LLM-based systems