AI Research Engineer, Pre-Training

Location: London, United Kingdom or New York, NY, United States

Salary: $250,000–$300,000 base per year, plus discretionary performance-based bonuses and benefits.

Company

hirify.global applies scientific methods and sophisticated computing infrastructure to algorithmic trading and develops foundation models for markets.

What you will do

• Improve large-scale pre-training systems for foundation models that process market and alternative data.
• Develop and optimize kernels using technologies such as CUDA, Triton, Pallas, and CuTe DSL.
• Improve training data loading, parallelism, networking, and fault tolerance.
• Develop lower-level components with PyTorch, JAX, XLA, CUDA Graphs, and related systems.
• Collaborate with researchers to co-design models and shape the research agenda.
• Optimize the use of large GPU clusters to maximize the impact of model training on trading.

Requirements

• Strong engineering skills in GPU systems, deep learning infrastructure, or related low-level technologies.
• At least two years of professional experience building deep learning systems.
• Experience with one or more of CUDA, Triton, Pallas, CuTe DSL, PyTorch, JAX, XLA, CUDA Graphs, FPGA, or ASIC development.
• Ability to translate methods between different application domains.

Nice to have

• Experience with large language models.
• Experience building deep learning systems in robotics, biology, chemistry, physics, audio, video, recommendations, or another domain.
• Finance experience is not required.

Culture & Benefits

• Scientific, research-driven approach to algorithmic trading.
• Access to a sophisticated computing environment with multiple modern GPU clusters.
• Openness, transparency, and collaboration across diverse areas of expertise.
• Competitive benefits package and performance-based bonuses.
• AI tools are prohibited during interviews or assessments unless explicitly authorized.