Quantitative AI/ML Engineer – Trading & Risk

Boundless; Dubai; On-site; AED 350,000 - 600,000; Full time

Job summary

Boundless is seeking a seasoned quantitative researcher to develop, validate and productionise ML models for understanding trading behaviour, execution quality and market exposure. You will turn research into reliable model services and collaborate with Dealing and Risk to ensure responsible use.

You will work with Python, SQL and time-series methods, handle real trading data, and contribute to model explainability, robustness and monitoring in a fast-paced brokerage environment.

Qualifications

• Strong Python, SQL, probability, statistics, optimisation and time-series analysis skills.
• Hands-on experience with scikit-learn, XGBoost, LightGBM or PyTorch.
• Experience with trading/execution data and risk analytics.
• Degree in a quantitative field; postgraduate study advantageous.

Responsibilities

• Translate approved research hypotheses into features, labels and models for behavioural classification, anomaly detection, exposure forecasting and execution-quality analysis.
• Develop models using trading activity, price movements, order timing, holding periods, execution data and account history, with clear definitions of what each prediction represents.
• Investigate concentration, correlated activity, changes in trading patterns and adverse‑selection indicators using supervised, unsupervised or time-series techniques as appropriate.
• Benchmark machine learning against transparent statistical and rules-based baselines; establish whether added complexity produces meaningful incremental value.
• Produce explainability, calibration, sensitivity and stability analyses; document assumptions, training populations, limitations and conditions under which the model should not be used.
• Package models for batch or online inference with versioned code, features, datasets and dependencies, working with MLOps on deployment and rollback.
• Implement model monitoring for data drift, prediction changes, performance deterioration and operational failures; distinguish changing markets from broken data.
• Run shadow or controlled challenger assessments before operational use, with approval and independent challenge for material trading or risk applications.
• Investigate model failures and refresh models through reviewed change procedures; support the Quantitative Lead in explaining results to Dealing and Risk.

Skills

Python

SQL

Time-series

Statistics

Probability

ML frameworks

Model deployment

Experiment tracking

Education

Bachelor's in Mathematics/Statistics/CS/Engineering

Postgraduate study advantageous

Tools

scikit-learn

XGBoost

LightGBM

PyTorch

Job description

Role purpose

Develop, validate and productionise machine-learning models that improve the brokerage’s understanding of trading behaviour, execution quality and market exposure. Apply rigorous statistical methods to noisy, time-dependent financial data and convert quantitative research into reliable model services. Work with the Quantitative Lead, Dealing and Risk to ensure outputs are interpretable and appropriate for their approved use.

Key responsibilities

Research and model development

• Translate approved research hypotheses into features, labels and models for behavioural classification, anomaly detection, exposure forecasting and execution-quality analysis.
• Develop models using trading activity, price movements, order timing, holding periods, execution data and account history, with clear definitions of what each prediction represents.
• Investigate concentration, correlated activity, changes in trading patterns and adverse-selection indicators using supervised, unsupervised or time-series techniques as appropriate.
• Benchmark machine learning against transparent statistical and rules-based baselines; establish whether added complexity produces meaningful incremental value.

Validation and interpretation

• Build chronological and walk-forward validation that reflects live information availability; prevent leakage from overlapping labels, related accounts and future market observations.
• Evaluate robustness across instruments, client cohorts and market regimes, accounting for class imbalance, multiple testing and uncertainty in small samples.
• Include realistic spreads, slippage, latency, liquidity constraints and transaction costs when evaluating execution or hedging-related proposals.
• Produce explainability, calibration, sensitivity and stability analyses; document assumptions, training populations, limitations and conditions under which the model should not be used.

Production models and monitoring

• Package models for batch or online inference with versioned code, features, datasets and dependencies, working with MLOps on deployment and rollback.
• Implement model monitoring for data drift, prediction changes, performance deterioration and operational failures; distinguish changing markets from broken data.
• Run shadow or controlled challenger assessments before operational use, with approval and independent challenge for material trading or risk applications.
• Investigate model failures and refresh models through reviewed change procedures; support the Quantitative Lead in explaining results to Dealing and Risk.

Essential experience and qualifications

• 4+ years in quantitative research, applied machine learning or financial data science, with experience implementing production models.
• Strong Python, SQL, probability, statistics, optimisation and time-series analysis skills.
• Hands-on experience with tools such as scikit-learn, XGBoost, LightGBM or PyTorch and a clear ability to select methods appropriate to the problem.
• Working knowledge of leveraged trading, FX/CFDs or comparable electronic markets, including exposure, leverage, margin and execution costs.
• Demonstrated ability to prevent data leakage, conduct out-of-sample validation and evaluate model stability under changing market conditions.
• Strong software engineering, experiment tracking and technical documentation skills; ability to communicate uncertainty rather than overstate predictive power.
• Degree in Mathematics, Statistics, Computer Science, Physics, Engineering or a related quantitative discipline; postgraduate study is advantageous.

Preferred experience

Direct experience with retail FX/CFD flow, broker risk books, hedging analytics or execution-quality modelling.

Experience with real-time inference, high-frequency data, graph methods or interpretable client segmentation.

Experience deploying models through monitored shadow testing and model review processes.