Machine Learning Engineer

About the Role

We’re looking for a Machine Learning Performance Engineer to make deep learning systems faster, more efficient, and easier to scale. You’ll work on the technology behind model training and inference, with a focus on GPU performance and low-level optimization.

Your work will support machine learning research used in quantitative trading. You’ll build training pipelines, improve open-source frameworks, and find ways to reduce computing costs and execution time.

This role is ideal for an engineer who enjoys working close to the hardware and understanding why software performs the way it does.

Key Responsibilities

Machine Learning Infrastructure

• Build reliable pipelines for deep learning training and inference.
• Improve the speed and efficiency of machine learning workloads.
• Develop infrastructure that supports large-scale model training.
• Reduce unnecessary memory usage and computing overhead.
• Work with researchers to improve how models run in production.

GPU Performance and Optimization

• Identify performance bottlenecks in GPU workloads.
• Develop and optimize CUDA kernels for deep learning operations.
• Improve memory access, parallel execution, and GPU resource usage.
• Use profiling tools to measure performance and guide improvements.
• Apply knowledge of GPU architecture to improve training and inference speed.
• Test changes to confirm performance gains without affecting correctness.

Deep Learning Framework Development

• Work with the internal components of PyTorch, JAX, or TensorFlow.
• Modify open-source frameworks to improve performance or add functionality.
• Investigate how frameworks manage computation, memory, and execution.
• Build tools that make machine learning experiments more efficient.
• Collaborate with engineers and researchers on technical improvements.

Required Skills and Qualifications

GPU Programming and Computer Architecture

• Strong experience with CUDA programming and GPU optimization.
• Deep understanding of GPU and CPU architecture.
• Knowledge of memory hierarchies, parallel processing, and hardware performance.
• Experience using CUDA Tensor Cores, cooperative groups, and CUDA Graphs.
• Familiarity with warp-level intrinsics and GPU execution behavior.
• Ability to profile code and identify performance limitations.

Programming and Machine Learning Frameworks

• C++: Strong skills in systems programming and performance-sensitive development.
• Python: Experience building machine learning tools and applications.
• PyTorch: Understanding of framework internals and execution behavior.
• JAX or TensorFlow: Experience working with deep learning frameworks at a technical level.
• Deep learning: Knowledge of training workflows, inference, and model execution.

You should be comfortable investigating framework internals rather than relying only on high-level APIs.

Additional Technical Experience

The following skills would be valuable:

• JAX ecosystem: Experience with XLA, Flax, or related tools.
• GPU libraries: Familiarity with Triton, CUB, cuDNN, or cuBLAS.
• Linux: Experience with Linux systems programming and performance tools.
• Distributed training: Knowledge of large-scale training across multiple GPUs or machines.
• Open-source development: Contributions to machine learning, scientific computing, or data science projects.

Engineering Environment

You’ll work with machine learning researchers and software engineers on systems that support data-driven trading strategies. The focus is on making models run faster while keeping results accurate and reliable.

Some projects will involve improving existing code. Others may require developing custom GPU kernels or changing how a framework handles computations.

What to Expect

• Low-level engineering: Work directly with GPU hardware concepts, CUDA, and framework internals.
• Performance testing: Measure results and use profiling data to guide improvements.
• Open-source technology: Work with established machine learning frameworks and supporting libraries.
• Research collaboration: Help researchers run experiments and train models more efficiently.
• Technical ownership: Investigate performance issues and develop solutions from start to finish.
• Financial applications: Learn how machine learning supports quantitative research and futures trading.

Compensation and Benefits

The position offers a competitive compensation package with performance-based rewards and employee benefits.

Compensation

• Competitive compensation based on experience and qualifications.
• Performance-based bonus opportunities.
• Participation in a global profit-sharing program.
• 401(k) retirement plan with an employer match of up to 50%, subject to plan terms.

Employee Benefits

• Medical, dental, and vision insurance.
• Mental health support and coverage.
• Disability and life insurance.
• 25 paid vacation days, plus market holidays.
• Office meals, including breakfast, lunch, and snacks.
• Employee events, clubs, and recreational activities.

Why Consider This Opportunity?

This Machine Learning Performance Engineer role offers the chance to solve challenging problems at the intersection of deep learning, GPU programming, and quantitative finance.

• Optimize CUDA workloads and improve GPU performance.
• Work with PyTorch, JAX, TensorFlow, and open-source ML tools.
• Build faster training and inference pipelines.
• Collaborate with researchers on demanding machine learning problems.
• Apply low-level programming skills to real computing challenges.

This opportunity is a strong fit for an engineer who enjoys performance optimization, understands computer architecture, and wants to improve the systems behind advanced machine learning applications.