AI Scientist (Financial World Model, Securities Trading)

Full Description

Company Description: A leading financial technology and securities firm with research hubs in Hong Kong and the United States, focusing on artificial intelligence applications in capital markets and quantitative trading.

Location: Hong Kong / United States / Remote

Nature: Permanent

Job Title: AI Scientist (Financial World Model, Securities Trading)

Responsibilities

• Develop advanced financial world models focused on securities trading and stock market structures

• Build simulation environments aligned with financial logic for forecasting and strategy testing

• Train probabilistic inference models using historical market data, trading rules, and unstructured information

• Research and optimize model architectures including Transformer, Mamba, MoE, Diffusion, and RLHF

• Collaborate with data engineering and quantitative teams on data cleaning, labeling, and transformation strategies

• Conduct performance analysis, reporting, experimental validation, and explainability studies of models

• Stay updated on cutting-edge research in financial AI, world models, large agents, and deep reinforcement learning

• Contribute to publications, technical papers, and knowledge sharing within the team

Requirements

• Hands-on experience in AI model training and deep learning frameworks such as PyTorch or JAX

• Strong mathematical foundation in probability, statistics, linear algebra, time series, and stochastic processes

• Proven experience in large-scale model training and big data processing

• Expertise in at least one domain: world models, probabilistic models, reinforcement learning, or generative models

• Excellent communication skills in both Chinese and English for cross-border collaboration

• Preferred background from leading AI teams such as Alibaba Qwen or DeepSeek core training teams

• Demonstrated achievements in world models, probabilistic forecasting, quantitative AI, or LLM training

• Familiarity with financial market microstructure, matching rules, and quantitative strategies

• Experience in distributed training frameworks such as DeepSpeed, Megatron, or ColossalAI

• Ability to work independently and deliver results in a fast-paced, international environment