Jobs at YNDX Family

Привет! Это сообщество Yandex Family и многие классные зарубежные ребята доверяют нам процессы по найму.Ищем Principal/Senior ML/Inference инженера в команду Mirai на полную удаленку и валютную зарплату.По зарплате готовы предлагать от $150,000 гросс в год.Mirai делает так, чтобы AI-модели работали быстро и дешево прямо на устройствах, а не в облаке. Строят high-performance inference engine под Apple Silicon и инфраструктуру вокруг него: профилирование/бенчмарки, runtime-оптимизации и т.д. Компания закрыла seed на $10M от топ VC и ангелов.Команда русскоязычная, маленькая, можно сильно влиять на продукт; фаундеры — создатели Reface (200M+ users, backed by a16z) и Prisma (100M+).About usMirai builds the fastest on-device inference engine for Apple Silicon. In under a year, a 14-person team built a full stack, from model optimization to a proprietary runtime, outperforming MLX and llama.cpp on supported models.We’re making local inference practical, fast, and reliable for real products.Why us?Mirai is founded by proven entrepreneurs who built and scaled consumer AI leaders like Reface (200M+ users, backed by Andreessen Horowitz) and Prisma (100M+ users). Our team is small (14 people), senior, and deeply technical. We ship fast and own problems end-to-end.We’re advised by a former Apple Distinguished Engineer who worked on MLX, and backed by leading AI-focused funds and individuals.ResponsibilitiesYou'll work across our inference engine and model conversion toolkit, implementing new model architectures, supporting new modalities, writing optimized kernels, and building a wide range of features such as function calling and batch decoding.This role is ideal for someone who reads papers for fun, enjoys writing high-performance code, and gets excited about constant learning.RequirementsJAX / Equinox / Pallas stack.Rust systems programming with a focus on developer experience.Writing Metal / Vulkan kernels.Neural codecs and voice model architectures.Trellis-based quantization approaches.Advanced speculative decoding methods, such as EAGLE.Deep understanding of Transformer / SSM / Diffusion / Vision language models.Benchmarking inference performance and model quality.Strong linear algebra, optimization methods, and probability theory.