Description
Jump Trading Group is hiring a Pre-Training Research Engineer to lead large-scale foundation model pre-training for ML-powered trading. The role focuses on building fault-tolerant distributed training infrastructure across thousands of GPUs and TPUs, engineering high-throughput data pipelines, designing custom kernels, and co-developing novel architectures and training techniques with researchers. The position requires strong expertise in Python, PyTorch or JAX, distributed systems, HPC, and published research in efficient ML training, with preferred experience in CUDA, Triton, hardware acceleration, and financial markets.
