株式会社Recho
Open positions
Senior LLM Training Engineer (Post-training)
Senior engineer owning LLM post-training for Recho's voice AI agents: design and run SFT, preference optimization and online RL, including reward design, training-data quality and in-house evaluation. Stack: Python, PyTorch, verl/OpenRLHF/TRL, vLLM/SGLang, FSDP/DeepSpeed. Tokyo hybrid, English-speaking team.
Senior Speech AI Engineer (Speech Synthesis / TTS)
Senior engineer who improves the quality of the voice behind Recho's phone-based AI customer-support agents: training state-of-the-art TTS models (PyTorch, LLM-based TTS, audio codecs, flow matching, diffusion), owning speech data strategy and evaluation, and partnering on latency — with a path to leading the speech team.
Senior AI Inference Engineer (LLM & Speech Models)
A hands-on senior engineer who designs, builds, and optimizes Recho's inference framework for speech models (ASR/TTS/speech-to-speech) and LLMs, owning throughput and GPU cost. Day to day: tuning vLLM/SGLang inference, dynamic batching for variable-length audio, async pipelines, and GPU orchestration with Kubernetes, Slurm, and Ray using Python and PyTorch.
SRE エンジニア
SRE エンジニア 仕事概要 ## 役割：システムの信頼性と安全性を最高水準へ導く 障害を減らすことはもちろん、障害発生時にいかに早く気づき、早くトリアージし、早く復旧できるかを重視しています。この一連の流れにおける人的関与を自動化し、運用工数を継続的に削減していくことを目指しています。 **■お任せしたい主な業務** **1. モニタリング・オブザーバビリティ** - Datadogによる各サーバーのo11y -…