Software Engineer – LLM Inference Optimization

Description

Preferred Networks (株式会社Preferred Networks) — Japan's leading AI unicorn, builder of the PLaMo LLM family and MN-Core AI processors, with research teams that have collaborated with Toyota, FANUC, and NTT on industrial AI — is hiring a Software Engineer to improve the inference engine powering its PLaMo API service and maintain PLaMo implementations in open-source projects including vLLM, based onsite in Tokyo, Japan. Preferred Networks will provide full visa sponsorship and relocation support for international candidates. This role focuses on optimizing how PLaMo models run in production — latency, throughput, cost — and contributing improvements upstream to open-source inference frameworks. It offers a rare opportunity to work at the frontier of Japanese-language AI while contributing meaningfully to the broader LLM inference ecosystem. The role was posted directly by Preferred Networks in the HN "Who is Hiring? (August 2026)" thread.

Required Skills