4 часа назад
Systems Engineer (AI)
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Описание вакансии
Текст:
TL;DR
Systems Engineer (AI): Developing and optimizing the Ollama local runtime for open models with an accent on GPU acceleration, memory management, and cross-platform hardware support. Focus on building high-performance inference engines in Go and C/C++ to ensure instant model execution across diverse consumer and enterprise hardware.
Company
is the leading developer network for open models, providing a local-first runtime used by over 8.9 million developers.
What you will do
- Optimize open model execution for speed and reliability across consumer and enterprise hardware.
- Own core runtime components including model loading, scheduling, memory management, and quantization.
- Integrate new model architectures and quantization formats to support the latest open models.
- Improve performance metrics such as cold-start times, time-to-first-token, and overall throughput.
- Collaborate with model labs and hardware vendors on deep integrations and early access.
- Contribute to the open-source community by shipping code directly to millions of users.
Requirements
- Location: Must be based in Palo Alto (On-site role)
- Strong systems fundamentals with proficiency in Go, C, or C++.
- Experience working close to the metal in areas like GPU compute, inference, game engines, databases, or OS development.
- Proven ability to profile and optimize real-world workloads for performance.
- Experience shipping software to large user bases across diverse hardware and OS combinations.
Nice to have
- Experience with model quantization techniques.
- Proficiency in GPU programming (CUDA, Metal, or SYCL).
- Experience with Apple MLX.
Culture & Benefits
- Small, talent-dense team with a flat, low-ego, and fast-moving culture.
- Design-driven environment focused on shipping high-impact code.
- Backed by top-tier investors including Y Combinator, Benchmark, 8VC, and Theory Ventures.
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 →