AI/ML Quant Researcher: Transformer Market Models

JPMorgan Chase & Co.; Greater London; On-site; GBP 120,000 - 180,000; Full time

Job summary

JPMorgan Chase & Co. in London seeks an AI/ML Quantitative Researcher to lead research on Transformer-based and time-series foundation models over large-scale market datasets, developing methods to make them robust, transferable, and measurable across instruments and regimes. The role requires hands-on experience pre-training large foundation models from scratch.

This position is designed for deep research with real constraints—scaling laws, data efficiency, and robustness are core to the agenda.

Qualifications

• A advanced degree in ML, CS, statistics or related quantitative field.
• Proven experience pre-training a large model from scratch.
• Experience building large-scale data pipelines and distributed training systems.
• Deep knowledge of optimization, parallelism, mixed precision, checkpoints, and benchmarking.
• Evidence of research quality through successful large-model training or deployments.

Responsibilities

• Pre-train Transformer-based and time-series foundation models from scratch using large-scale market datasets.
• Develop data representations, tokenization, self-supervised objectives, and distributed training recipes.
• Fine-tune and post-train models for alpha generation, pricing, market making, execution, and risk tasks.
• Study scaling laws, transfer across instruments, regime robustness, data efficiency, and trade-offs of model quality vs. latency.
• Design evaluation protocols linking pre-training metrics to economically meaningful outcomes.
• Build reusable training, checkpointing, evaluation, and model-serving components with ML infrastructure engineers.

Skills

Pre-training large models from scratch

PyTorch / JAX / distributed training

Large-model training & evaluation

Education

Master's or PhD in ML / CS / statistics

Tools

PyTorch

JAX

Job description

JPMorgan Chase & Co. in London seeks an AI/ML Quantitative Researcher to lead research on Transformer-based and time-series foundation models over large-scale market datasets, developing methods to make them robust, transferable, and measurable across instruments and regimes. The role requires hands-on experience pre-training large foundation models from scratch.

This position is designed for deep research with real constraints—scaling laws, data efficiency, and robustness are core to the agenda.