Вакансия из Hirify Global, списка международных tech-компаний Для мэтча и отклика нужен Plus
Пожаловаться
Зарплата и рынок
развернуть
$165к-300к/год
$140к$180кмедиана$210к
По похожим ролям (49) ~ $180к/год
На СНГ рынке: $2.7к/мес ($1.8к - $3.6к)
91
Крутая вакансия
развернуть
У роли чёткий фокус на производительности GPU, серьёзная техническая глубина и участие в масштабном переходе на ускоренные вычисления - это сильная возможность для развития в узкой инженерной специализации.
Передовые технологии
Чёткие задачи
Важная инициатива
Поддержка обучения
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Подключить Plus
Описание вакансии
Текст:
Hirify AI/Оригинал
TL;DR
GPU Performance Engineer (CUDA): Building and optimizing GPU-accelerated kernels, libraries, and abstractions for financial computation workloads with an accent on CUDA programming, memory efficiency, and mixed-precision computation. Focus on profiling workloads with NVIDIA tooling, optimizing performance across next-generation hardware, and integrating multi-GPU libraries for quantitative modeling.
Location: Hybrid in NY, New York, United States, with flexible in-office days
Salary: $165,000–$300,000 base pay annually, plus potential discretionary bonus and benefits
Company
hirify.global is a quantitative investment management and trading firm applying technology, artificial intelligence, data science, and quantitative research to financial markets.
What you will do
Design and implement GPU-accelerated kernels for financial computation workloads.
Optimize GPU code for throughput, latency, and memory efficiency across current and next-generation hardware, including Blackwell and Rubin.
Develop precision-management procedures for FP8/FP4 training and inference in financial applications.
Profile and optimize workloads using NVIDIA Nsight Systems and Nsight Compute.
Build reusable GPU libraries and abstractions for modeling teams.
Evaluate and integrate RAPIDS, CUTLASS, cuBLAS, and TensorRT for financial use cases.
Requirements
BS or MS in Science, Technology, Engineering, or Math.
At least 1 year of experience required; 4–10 years preferred.
Expert-level CUDA programming, including kernel development, memory management, and stream and graph optimization.
Deep understanding of GPU architecture, including SM structure, warp scheduling, memory hierarchy, registers, shared memory, L1/L2, and HBM.
Experience profiling and optimizing GPU workloads.
Strong C++ and Python skills, with familiarity with mixed-precision computation and numerical stability.
Demonstrated ability to deliver meaningful speedups on real workloads rather than only benchmarks.
Nice to have
Background in HPC, scientific computing, or computational finance.
Experience with multi-GPU and multi-node GPU programming using NCCL or MPI.
Familiarity with GPU-accelerated data-processing frameworks such as RAPIDS and cuDF.
Culture & Benefits
Hybrid work policy with flexible in-office days and a home-office setup budget.
Fully paid medical and dental insurance for employees and dependents, plus employer-paid life and disability insurance.
Tuition reimbursement and sponsorship for conferences and training.
Onsite gyms with laundry service, wellness activities, casual dress, snacks, and game rooms.
Будьте осторожны
Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →