AI Research Engineer, Pre-Training

About The Opportunity

A leading quantitative financial technology enterprise is seeking an exceptional AI Research Engineer to advance its large-scale pre-training capabilities. In this role, you will help architect, optimize, and deploy massive foundation models designed to process extensive market and alternative data streams. If you want your engineering contributions to directly drive high-impact trading outcomes while leveraging extensive GPU infrastructure, this role offers an ideal environment with zero prior financial background required.

Responsibilities

• Optimize all phases of distributed deep learning pre-training, focusing on networking performance, memory management, data pipelines, and fault-tolerant execution.
• Collaborate directly with machine learning researchers to co-design network architectures and establish long-term compute roadmaps.
• Write high-performance lower-level code and custom kernels to maximize compute utilization across large-scale GPU infrastructure.
• Adapt and translate cutting-edge deep learning methodologies from non-financial domains into production-grade trading models.

Requirements

Must-have:

• 2+ years of hands-on professional experience engineering deep learning systems across any technical field (e.g., robotics, physical sciences, computer vision, audio, or recommendation systems).
• Deep expertise in modern hardware acceleration and deep learning frameworks (such as PyTorch, JAX, CUDA, Triton, or specialized compiler DSLs).
• Proven track record of developing or tuning low-level training infrastructure, custom kernels, or distributed training workflows (e.g., parallelism techniques, CUDA Graphs, XLA).
• Ability to solve open-ended systems performance challenges without relying on off-the-shelf software packages.
• No prior experience in finance or quantitative trading is necessary.

Preferred Qualifications

• Practical experience building or fine-tuning Large Language Models (LLMs) at scale.
• Background in specialized hardware programming, FPGA/ASIC integration, or custom compute acceleration.
• Strong track record of cross-domain methodology transfer.

Compensation & Benefits

• Highly competitive base salary and generous performance-based bonuses.
• Comprehensive benefits.
• Access to exceptional GPU-per-engineer ratios and state-of-the-art compute resources.

Location: New York, NY or London, UK