Apr 2025 — Present
SRE Engineer
Rakuten Travel
Managing production infrastructure for one of Japan's largest travel platforms, with site reliability engineering work for high-availability systems.
Experience
Engineering across production infrastructure, AI robotics, and product teams.
Internship
Solutions Engineer Intern
Osaro, Inc.
Worked on AI-powered robotic systems for warehouse automation, collaborating across engineering and customer-facing teams to deploy AI solutions in real-world environments.
Projects
Products shipped beyond the prototype stage.
BuddhaChat
AI Buddhist Counselor
An iOS app providing compassionate, Buddhist-inspired guidance powered by Google's Gemini API. Built with SwiftUI, GCP Cloud Run, Firebase, Stripe subscriptions, and Meta SDK campaign integration.
Swift, SwiftUI, Gemini API, GCP Cloud Run, Cloud Functions, Firebase, Stripe
ReportAI
AI Writing Assistant for Students
A full-stack SaaS product that helps university students draft academic reports with AI, including user authentication and payment processing.
React, Next.js, Node.js, Firebase, Stripe
tsukusuta
Custom LINE Sticker Maker
A web app that turned user-uploaded images into LINE-ready sticker sets. AI-driven image processing handled background removal and per-sticker formatting, with Cloud Functions automating the upload-to-publish pipeline on Firebase.
Next.js, TypeScript, Firebase Auth, Firestore, Cloud Functions, AI image processing
Research
Peer-reviewed robot learning work, published as an undergraduate.
Jan 2025
IEEE/SICE SII 2025
Curriculum Reinforcement Learning for Obstacle Avoidance Postures for a Hyper-redundant Manipulator
Undergraduate thesis work at the Complex Robot Systems Lab, Meiji University. Hyper-redundant manipulators are well suited to cluttered, narrow workspaces, but the same extra degrees of freedom make them hard to control: analytical methods scale poorly with joint count and get stuck in local minima, while naive deep RL sees its exploration diverge in the high-dimensional action space. I trained the policy with PPO under a curriculum that tightens obstacle and goal conditions stage by stage, paired with a reward balancing collision avoidance against goal reaching — improving both learning efficiency and avoidance performance over training without a curriculum.
Reinforcement learning, Curriculum learning, PPO, Hyper-redundant manipulator, Obstacle avoidance, Motion planning, Python, PyTorch
About
I bridge cutting-edge AI and real-world product experiences.
Tokyo-based software engineer working at the intersection of AI and product. Graduated from Meiji University, Faculty of Science and Technology, where my undergraduate research on curriculum reinforcement learning for hyper-redundant manipulators was published and presented at IEEE/SICE SII 2025 in Munich.
SRE at Rakuten Travel by day, indie maker of AI apps by night. Previously worked on AI-powered robotics at Osaro, Inc. in the U.S. Bilingual in Japanese and English.
Skills
A stack shaped by shipping AI products end to end.
Languages
Swift, Python, JavaScript, TypeScript, Node.js, SQL
Frontend
SwiftUI, React, Next.js
Backend & Infra
GCP Cloud Run, Cloud Functions, Firebase, Firestore, Docker, Nginx
Data & APIs
Gemini API, OpenAI API, Stripe API, X API, Meta SDK
DevOps & Tools
Git, Linux, Ubuntu, tmux, FFmpeg, MySQL
Contact