Performance Engineer

TL;DR

Performance Engineer (PyTorch/GPU): Analyze and optimize systems enabling trading strategies at scale with an accent on infrastructure efficiency across Python, PyTorch, OS, networking, storage, and CPU/GPU layers. Focus on identifying optimization opportunities, implementing performance telemetry, and scaling core models to free up compute capacity and reduce costs.

Location: Hybrid in New York City office (minimum 3 days a week)

Company

Quantitative investment manager with 30+ year track record of excellence, high employee retention, and a close-knit, diverse community.

What you will do

• Analyze system performance to enhance researcher throughput and velocity.
• Optimize compute resource utilization across software, systems, and hardware layers.
• Read codebases, suggest optimizations, and contribute PRs.
• Free up capacity, reduce costs, and support scaling of core models.
• Propose and implement systems for performance telemetry.
• Conduct PoC evaluations and contribute to system design.

Requirements

• Hybrid work in New York City office minimum 3 days/week
• Strong proficiency in Linux and performance engineering toolset.
• Experience with PyTorch, GPUs, and CUDA for optimization.
• Deep understanding of hardware-software interface.
• Versatile engineering mindset with coding, micro-optimization, and multi-language skills.
• Ability to analyze performance holistically.