田中 悠真
Tanaka Yūma
Senior Software Engineer · 12 Years of Experience
シニアソフトウェアエンジニア — 12年の経験
Tokyo, Japan
Download CV自己紹介
About
Based in Tokyo, I've spent 12 years building at the intersection of AI/ML and full-stack engineering. From deploying large language models in production to designing real-time inference pipelines and intuitive frontend experiences, I'm passionate about shipping AI products end-to-end.
技術力
Skills & Technologies
職歴
Career History
Preferred Networks, Inc.
株式会社プリファードネットワークス
Staff AI Full Stack Engineer / Tech Lead
- —Built an enterprise RAG platform processing 10M+ documents with LangChain + React conversational UI
- —Architected real-time inference serving on Kubernetes — 50K req/s, sub-200ms P99 latency
- —Created internal model evaluation framework, cutting LLM deployment validation from days to hours
- —Led cross-functional team of 6 ML engineers and 4 frontend engineers
LINE Corporation
LINE株式会社
Senior AI Engineer
- —Shipped smart reply feature generating contextual suggestions for 200M+ users
- —Built ML feature store and real-time prediction pipeline (Kafka → Spark → Redis)
- —Developed annotation platform UI (React + D3.js) used by 50+ internal labelers, 3x throughput improvement
Recruit Holdings Co., Ltd.
株式会社リクルート
ML / Full Stack Engineer
- —Built NLP-powered job matching algorithms for Indeed Japan, +25% match rate
- —Developed recruiter dashboard (Vue.js + FastAPI) visualizing ML predictions
- —Implemented automated model retraining pipelines with Airflow and MLflow
Rakuten Group, Inc.
楽天グループ株式会社
Software Engineer
- —Developed product recommendation APIs using collaborative filtering, 100K+ RPM
- —Built real-time search autocomplete with Elasticsearch and behavioral signals
- —Migrated legacy product pages to React SPA with server-side rendering
作品集
Projects
Kotoba
Production RAG Framework
A lightweight Python framework for building production-grade retrieval-augmented generation apps. Supports pluggable retrievers, rerankers, and any LLM provider out of the box.
Sensei
AI Code Review Bot
A GitHub App that provides context-aware, team-style-aligned code review comments using fine-tuned LLMs. Learns your team's conventions and catches bugs before humans do.
Miru
AI Annotation Platform
Full-stack annotation tool for image and video ML datasets with AI-assisted pre-labeling. Supports bounding boxes, segmentation masks, and keypoints with real-time collaboration.
Sakura
AI-First Component Library
A React component library with 60+ components including AI-specific UI patterns — streaming chat bubbles, confidence meters, model selectors, and token usage displays.
Nagare
Real-time ML Observability
Monitoring dashboard that visualizes ML pipeline health, model drift detection, data quality scores, and inference latency. Reduced incident detection time by 45%.
Tsuki
Smart Finance Tracker
Mobile-first PWA for personal expense tracking with AI-powered category suggestions, receipt OCR scanning, and monthly spending insights generated by LLM analysis.
学歴・資格
Education & Certifications
学歴
東京大学
The University of Tokyo
M.Eng. in Computer Science — NLP Lab
Thesis: “Attention-Based Neural Architectures for Cross-Lingual Document Summarization”
京都大学
Kyoto University
B.Eng. in Informatics
資格
AWS Machine Learning — Specialty
GCP Professional ML Engineer
NVIDIA DLI — Fundamentals of Deep Learning
Registered Information Security Specialist (情報処理安全確保支援士)
哲学
Philosophy
「AIは道具、人が主役」
“AI is the tool — people are the protagonist.”
I believe the best AI products disappear into the workflow. Great engineering means building systems that amplify human judgment, not replace it — with reliability, transparency, and continuous improvement at every layer of the stack.
お問い合わせ
Contact
Let's connect
お気軽にご連絡ください