Machine Learning Researcher

AAA Global; Hong Kong; On-site; HKD 1,000,000 - 1,500,000; Full time

Job summary

AAA Global is seeking a senior research professional in Hong Kong to design and own deep learning pipelines for quantitative alpha research. You will drive the research agenda end-to-end, from problem formulation to deployment, while upholding rigorous out-of-sample evaluation and honest benchmarking against baselines.

The role emphasizes leadership in architecture selection, training diagnostics, and cross-team collaboration, with opportunities to influence the research direction across the

Qualifications

• Advanced degree in a quantitative field; strong academic record.
• 3–5 years applying deep learning to large-scale problems, ideally in finance.
• Proven end-to-end ownership of a DL model lifecycle on production or research lines.
• Strong Python programming and modern DL framework experience.
• Experience with large-scale model training: distributed/multi-GPU, mixed precision.
• Solid foundations in statistics, optimization, and ML.
• Familiarity with modern DL architectures and selection judgment.
• Practical techniques for low signal-to-noise learning and robust validation.
• Experience with large-scale datasets and efficient data pipelines.
• Fluency with experiment-management tooling and reproducible research environments.
• Knowledge of C++/CUDA and LLM tooling is a plus.

Responsibilities

• Design and build core DL pipelines for quantitative alpha research, from data prep to production deployment.
• Own a significant portion of the research agenda, including problem formulation, model design, training, and evaluation.
• Maintain strict out-of-sample discipline and robust benchmarking against baselines.
• Advise on architecture choices and training diagnostics; review model designs.
• Create standardized training/inference interfaces and reusable components for cross-team use.

Skills

Python

Deep Learning

Distributed Training

Statistics

Optimization

C++/CUDA

Experiment Tracking

LLM Tooling

Education

PhD or MSc in CS/Math/Physics/Statistics

Strong academic record

Tools

PyTorch

CUDA

TensorFlow

Profiling tools

Job description

We are helping a top tier quantitative hedge fund looking for best talent to build and own the deep learning engine behind the firm's alpha research. You will design the core pipelines, drive a meaningful share of the research agenda end-to-end, and act as a central point of deep learning expertise across the organization. It is a high-impact, high-autonomy role for a researcher who pairs strong applied modeling with genuine scientific rigor.

Responsibilities

• Design and build the firm's core deep learning pipelines for applied quantitative alpha research — from data preparation and distributed training through evaluation and production deployment.
• Own a significant part of the research agenda, running the full empirical loop: problem formulation, model design, training, validation, and performance attribution.
• Set the standard for research discipline in a low signal-to-noise domain — strict out-of-sample hygiene, leakage prevention, and honest benchmarking against simpler baselines.
• Serve as a go-to source of deep learning expertise: advise on architecture selection and training diagnostics, review model designs, and shape how models are evaluated and promoted.
• Build standardized training and inference interfaces and reusable components that let models flow smoothly across teams and systems.

Requirements

• An advanced degree in Computer Science, Engineering, Physics, Mathematics, Statistics, or a related quantitative field, with a strong academic record from a leading university. A background of distinction in mathematics or science competitions (e.g., IMO, IOI, IPhO and national equivalents) is a plus.
• Around 3–5 years applying deep learning to large-scale problems, ideally in quantitative finance. A strong PhD research record combined with hands‑on experience training large models at a leading AI/technology company is equally welcome, and prior experience at a top quantitative trading firm is highly valued.
• Proven end-to-end ownership of the deep learning model lifecycle on at least one significant production system or research line.
• Strong programming skills in Python and a modern deep learning framework.
• Hands‑on experience with large-scale model training: distributed/multi‑GPU training, mixed precision, and throughput profiling and optimization.
• Solid foundations in statistics, optimization, and machine learning.
• Command of modern deep learning architectures — and the judgment to know when a simpler model should win.
• Practical technique for low signal-to-noise learning: regularization, ensembling, and validation protocols that survive out-of-sample.
• Experience with large-scale datasets — efficient columnar formats, streaming data loaders, and point‑in‑time‑correct dataset construction.
• Fluency with experiment‑management tooling: experiment tracking, hyperparameter optimization, and reproducible research environments.
• Working knowledge of C++ or CUDA-level optimization, and familiarity with LLM tooling as a research accelerant, are a plus.