Machine Learning Engineer – Deep Learning for High-Frequency Trading

About the Company

Our client is a well-established quantitative trading firm specializing in artificial intelligence and machine learning applications for high-frequency trading (HFT) and market making.

Job Overview

This position reports directly to the Head of HFT and involves developing and deploying large-scale neural networks trained with raw tick-level order book data from multiple markets and exchanges. The role includes full ownership of data preparation, model development, training infrastructure, and seamless integration into a live C++ trading platform. The team operates with minimal hierarchical structure, granting direct communication with key decision-makers and significant responsibility over your projects. Competitive market compensation includes potential upside linked to profit and loss performance.

Key Responsibilities

• Conduct research and train deep learning models on extensive order-book and market datasets to forecast short-term market movements.
• Develop and experiment with sequence and time-series models tailored to handle noisy and rapidly changing financial data.
• Manage the complete data pipeline including preparation, feature and label engineering, and maintaining reproducible training experiments.
• Assess model performance with rigorous time-aware out-of-sample validation techniques, ensuring no data leakage or overfitting occurs.
• Enhance training scalability and efficiency by leveraging GPU resources, distributed computing, and performance tuning.
• Collaborate with HFT researchers and engineers to optimize model inference and integrate models into a latency-sensitive C++ market-making system.
• Monitor production model outputs and apply live feedback loops to direct further research and improvements.

Candidate Requirements

• A Master’s or PhD degree in Computer Science, Machine Learning, Mathematics, Physics, Electrical Engineering, or a related quantitative field; or equivalent significant research and engineering capabilities.
• Between 3 to 10 years of relevant experience post-degree; exceptional recent PhD graduates with robust hands-on training and implementation experience are also encouraged to apply.
• Proven ownership and experience training deep learning models at scale, handling datasets larger than 100 GB and utilizing multi-GPU environments.
• Proficient in Python with extensive use of PyTorch or JAX, alongside skills in C++, Rust, or CUDA.
• Strong practical experience in developing, implementing, and optimizing deep learning models, particularly under latency or throughput constraints; recent PhD holders may demonstrate suitability through research projects, benchmarking exercises, or internships.
• Documented track record in rigorous experimental research, evidenced by first-author publications at top machine learning conferences, Kaggle Master or higher ranking, or solid industry research with credible benchmarks.
• Excellent mathematical and statistical insight with the ability to design robust experiments, interpret outcomes accurately, and evaluate model generalization.
• Relevant experience with sequence modeling, time-series forecasting, or efficient deep learning architectures is highly valued.
• Additional advantages include expertise in custom CUDA or Triton kernel development, distributed training techniques, inference optimization, and knowledge of order-book dynamics and market microstructure.
• Genuine interest in applying deep learning methods to financial markets; prior trading experience is beneficial but not mandatory.

Application Details

Interested applicants should send their resume to Tina Wang via email (tw@empowerpartners.sg) with the subject line “Machine Learning Engineer – Deep Learning for HFT.” Please include relevant links to Google Scholar, GitHub, or Kaggle profiles if available. Only those shortlisted will be contacted.

Regulatory Information

License Number: 24S2395
Registration Number: R2090553