Вакансия из Hirify Global, списка международных tech-компаний Для мэтча и отклика нужен Plus
Пожаловаться
Зарплата и рынок
развернуть
$176к-243к/год
$180к$220кмедиана$240к
По похожим ролям (48) ~ $220к/год
На СНГ рынке: $4.4к/мес ($3.6к - $5.4к)
91
Крутая вакансия
развернуть
Чётко очерченная роль на стыке передового ИИ и кибербезопасности даёт серьёзную техническую глубину, но ограничение по США сужает круг тех, кому она доступна.
Только для США
Передовой стек
На стыке ИИ и безопасности
Чёткие задачи
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Подключить Plus
Описание вакансии
Текст:
Hirify AI/Оригинал
TL;DR
Reinforcement Learning Engineer (AI) (RLVR/cybersecurity): Building verifiable reinforcement learning environments, automated vulnerability verification pipelines, and scalable execution sandboxes for LLM agents with an accent on deterministic reward functions, exploitation, fuzzing, and low-level program analysis. Focus on designing reproducible environments, preventing reward hacking, instrumenting system behavior, and optimizing massive-scale parallel evaluation and distributed RL workflows.
Location: Remote - US; remote work-from-home 100% of the time.
Salary: $176,400–$242,550 base range, with possible discretionary bonus or commission eligibility.
Company
hirify.global is a cybersecurity platform combining exposure discovery, offensive testing, AI, and human insight to help organizations identify and prevent real-world security risks.
What you will do
Design, build, and deploy high-throughput RLVR environments that evaluate LLM action sequences against deterministic execution outcomes.
Develop automated test harnesses, sandboxes, verification engines, and binary pass/fail reward signals for memory corruption, web security, and logic vulnerabilities.
Integrate the Mayhem automated analysis platform and real-world vulnerability feeds into scalable environment-generation pipelines.
Architect safe, isolated, reproducible Docker, BuildKit, or Nix environments for simultaneous agent-driven exploitation and patching trajectories.
Collaborate with researchers at Anthropic, OpenAI, and Cohere on benchmark formats, observation spaces, and verifiable cybersecurity evaluation metrics.
Implement telemetry, ground-truth verification, trajectory logging, low-level instrumentation, debugging tools, and performance optimizations for large-scale distributed RL workflows.
Requirements
Understanding of modern LLM RL training workflows, execution-based feedback, or Reinforcement Learning from Verifiable Rewards (RLVR).
Proficiency developing applications in Python and low-level systems programming in C.
Solid understanding of software vulnerabilities, binary exploitation, fuzzing methodologies, or program analysis.
Experience with DevOps pipelines such as GitHub Actions, reproducible builds using Docker, BuildKit, or Nix, Linux systems, and low-level debugging.
Experience working with or building benchmark environments such as CTFs, SWE-bench, security challenges, or execution sandboxes.
Ability to complete the physical requirements of the role with or without reasonable accommodation, including remaining stationary 50% of the time and carrying or moving a laptop as needed.
Nice to have
Rust experience.
Experience designing custom reward functions, ground-truth verifiers, or automated grading engines for AI safety and reasoning models.
Background with low-level program analysis tools, sanitizers such as ASan or MSan, compiler instrumentation, or automated exploit generation tools.
Experience participating in or developing competitive cybersecurity benchmarks, CTFs, or open-source AI evaluation frameworks.
Culture & Benefits
Remote, work-from-home arrangement 100% of the time.
Inclusive workplace focused on diversity, belonging, and reasonable accommodations.
Collaborative environment described as a family-oriented culture.
Background checks may be conducted because the role involves access to confidential and proprietary information.
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 →