Research Engineer (AI)
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Описание вакансии
TL;DR
Research Engineer (AI): Developing and optimizing large-scale model architectures and internet-scale data pipelines with an accent on inference throughput, model training, and distributed systems. Focus on pushing the boundaries of long-context LLMs, reinforcement learning, and deploying AGI safely through hands-on technical contribution.
Location: Onsite in San Francisco, with relocation support provided.
Salary: $225,000–$550,000
Company
is a fast-paced, research-driven startup focused on the safe development and deployment of AGI.
What you will do
- Optimize inference throughput for novel model architectures to improve system efficiency.
- Train trillion-parameter models on large-scale GPU clusters.
- Design, prototype, and optimize new model architectures from the ground up.
- Build and maintain internet-scale data pipelines and web crawlers.
- Curate high-quality post-training datasets to enhance model capabilities.
- Contribute to research across long-context, inference-time compute, and reinforcement learning.
Requirements
- Strong general software engineering background.
- Deep knowledge of current deep learning literature.
- Hands-on experience with pre- and post-training of large language models.
- Proven ability to ideate, prototype, and evaluate research concepts.
- Experience designing and managing large-scale distributed systems and ETL workloads.
Culture & Benefits
- Comprehensive 401(k) plan with 6% company matching.
- Full health, dental, and vision insurance coverage for employees and dependents.
- Unlimited paid time off policy.
- Significant equity participation as part of total compensation.
- Visa sponsorship and relocation stipend available for San Francisco.
- Focused, small-team environment centered on integrity, quality, and hands-on collaboration.
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 →