Design and build end-to-end infrastructure for training, evaluation, and productionization of ML models, working closely with our HPC engineers who manage our on-prem compute cluster
Influence foundational choices around data access, compute orchestration, experiment tracking, model versioning, and deployment pipelines
Partner with quant researchers to accelerate iteration cycles, tighten feedback loops, and bring models from prototype to live trading
Work with researchers to adapt and deploy modern architectures — transformers, state-space models, temporal convolutions, graph neural networks — to noisy, high-frequency financial data. Explore techniques like self-supervised pretraining, representation learning, and cross-sectional modelling where they offer genuine edge
Shape our approach to reproducibility, continual learning, and production monitoring across a petabyte-scale data environment
Define standards that create consistency across teams and geographies; mentor engineers and influence technical culture beyond your immediate work
Keep pace with developments in deep learning research and ML infrastructure; bring ideas from academia and industry into how we work — whether that's new architectures, training techniques, or tooling

8+ years of experience building ML platforms or infrastructure at a leading tech company, research lab, or quantitative firm
A track record of designing and owning large-scale training and inference systems — not just contributing, but architecting
Deep proficiency in Python, with strong experience in either CUDA or C++
Hands-on expertise with modern deep learning frameworks (PyTorch, TensorFlow, or JAX) and practical experience implementing architectures like transformers, attention mechanisms, or sequence models
Strong foundation in deep learning fundamentals: optimization, regularization, loss design, and the trade-offs that matter when training at scale
Experience with distributed training at scale (Horovod, NCCL) and GPU optimization (cuDNN, TensorRT)
History of deploying models to production with strong observability, reproducibility, and monitoring practices
Comfort working across the ML stack from data pipelines to training infrastructure to serving systems

Build, don't inherit — You'll make foundational technology choices in a platform that's still being defined, not maintain someone else's legacy
Real investment, real backing — This is a strategic priority with resources behind it, not a side experiment
Direct impact on trading — Your infrastructure will power models that make real trading decisions in competitive global markets
Global scope — Work with teams across New York, Chicago, Amsterdam, London, Sydney, Hong Kong and beyond; define practices that can scale worldwide
Ideas over titles — IMC's culture values clarity, rigor, and collaboration. The best ideas win, regardless of where they come from
Tight coupling with research — You won't be building in isolation. Researchers and engineers work side-by-side, iterating together

IMC is hiring a Principal Machine Learning Engineer