会社名

楽天グループ株式会社

会社概要

未来を信じ、より良い明日を創っていく。
イノベーションを通じて、人々と社会をエンパワーメントする。私たちは、そんな想いを大切に世界の人々に喜びと楽しさを届けます。
楽天は、E コマース、FinTech、デジタルコンテンツ、通信など、70 を超えるサービスを展開し、世界10 億以上のユーザーに利用されています。
これら様々なサービスを、楽天会員を中心としたメンバーシップを軸に有機的に結び付け、他にはない独自の「楽天エコシステム」を形成しています。ダイバーシティ推進は、楽天にとって最優先の企業戦略のひとつです。従業員の出身は70カ国・地域以上。世界中からユニークで多様な文化的背景や視点を持つ優秀な人材が集まり、イノベーションの原動力になっています。社内カフェテリアにはベジタリアン、ハラル対応のメニューを用意。礼拝所（Prayer room）もあります。
また、仕事と育児の両立支援や、障がい者雇用・活躍促進も積極的に推進。社内のLGBT（※1）当事者やアライ（※2）に対して、情報共有やサポート体制の強化も進めています。誰もが自分らしく力を最大限発揮して働ける。それが楽天のダイバーシティです。

70を超えるサービスを提供し、世界30カ国にサービス展開拠点を持ち、従業員の出身国・地域数は100を超え、オープンポジション制度を活用して多様なキャリアを描くことができる点も魅力です。
フレックスタイム制度、事情に応じたリモートワークの活用が可能です。本社には託児所やフィットネスジム、三食無料で利用可能なカフェテリアが併設されるなど、社員を支える環境が整備されています。

ポジション

1027504　Distributed Training & Inference Optimization Engineer (LLM) - GPU Optimization Department (GPUOD)

仕事内容

Department Overview
GPU Optimization Department is responsible for the strategic management, optimization, and governance of Rakuten's company-wide AI infrastructure, ensuring high-performance, cost-efficient utilization of compute resources for machine learning workloads. We oversee a large-scale hybrid infrastructure spanning thousands of accelerators, including the latest Hopper and upcoming Blackwell architectures.
As a central enabler for AI innovation, we:
- Optimize compute resource allocation across on-premises and multi-cloud environments, maximizing efficiency for training and inference workloads
- Manage hybrid orchestration of diverse accelerator resources, ensuring seamless scalability and cost-effective deployment
- Develop and enhance frameworks for large-scale distributed training, with special focus on LLMs and generative AI
- Optimize inference performance through model optimization techniques and system-level acceleration
- Collaborate with internal teams to deliver scalable, high-availability inference services tailored to business needs
- Continuously evaluate next-generation hardware solutions, including specialized AI chips optimized for LLM workloads
- By effectively managing both conventional and specialized compute resources across on-premises and cloud environments, our team ensures Rakuten's AI ecosystem remains at the forefront of performance, reliability, and cost-efficiency.

　
Position Details
As a GPU Training & Inference Optimization Engineer, you will focus on maximizing the performance, efficiency, and scalability of LLM training and inference workloads on Rakuten’s GPU clusters. You will deeply optimize training frameworks (e.g., PyTorch, DeepSpeed, FSDP) and inference engines (e.g., vLLM, TensorRT-LLM, Triton, SGLang), ensuring Rakuten’s AI models run at peak efficiency.
This role requires strong expertise in GPU-accelerated ML frameworks, distributed training, and inference optimization, with a focus on reducing training time, improving GPU utilization, and minimizing inference latency.

Key Responsibilities
- Optimize LLM training frameworks (e.g., PyTorch, DeepSpeed, Megatron-LM, FSDP) to maximize GPU utilization and reduce training time.
- Profile and optimize distributed training bottlenecks (e.g., NCCL issues, CUDA kernel efficiency, communication overhead).
- Implement and tune inference optimizations (e.g., quantization, dynamic batching, KV caching) for low-latency, high-throughput LLM serving (vLLM, TensorRT-LLM, Triton, SGLang).
- Collaborate with infrastructure teams to improve GPU cluster scheduling, resource allocation, and fault tolerance for large-scale training jobs.
- Develop benchmarking tools to measure and improve training throughput, memory efficiency, and inference latency.
- Research and apply cutting-edge techniques (e.g., mixture-of-experts, speculative decoding) to optimize LLM performance.

求める経験・スキル

Mandatory Qualifications:
- 3+ years of hands-on experience in GPU-accelerated ML training & inference optimization, preferably for LLMs or large-scale deep learning models.
- Deep expertise in PyTorch, DeepSpeed, FSDP, or Megatron-LM, with experience in distributed training optimizations.
- Strong knowledge of LLM inference optimizations (e.g., quantization, pruning, KV caching, continuous batching).
- Bachelor’s or higher degree in Computer Science, Engineering, or related field.
　
Desired Qualifications:
- Proficiency in CUDA, Triton kernel, NVIDIA tools (Nsight, NCCL), and performance profiling (e.g., PyTorch Profiler, TensorBoard).
- Experience with LLM-specific optimizations (e.g., FlashAttention, PagedAttention, LoRA, speculative decoding).
- Familiarity with Kubernetes (K8s) for GPU workloads (e.g., KubeFlow, Volcano).
- Contributions to open-source ML frameworks (e.g., PyTorch, DeepSpeed, vLLM).
- Experience with inference serving frameworks (e.g., vLLM, TensorRT-LLM, Triton, Hugging Face TGI).