01Coding × Design × Photography

Ray Otsuka大塚 嶺

02Approach

On building things actually used in the real world.

I've been programming since I was eleven, anchored to one idea: technology that makes daily life a little more fun. In high school I built AIbou, an AI English-conversation app that integrated GPT-3 before its public release — over 1,800 downloads, modest revenue, and a lot of feedback collected by handing posters out around Shibuya.

At university I led the design and build of a long-horizon memory function for an AI voice companion for elderly users. A graph-theoretic water-flow model decides not just what to remember, but how strongly — modeling gradual forgetting and the integration of episodic into semantic memory. In production today; patent pending.

There are plenty of engineers with stronger fundamentals and plenty of people who understand graph theory better than I do. Fewer people, I think, take that knowledge all the way through to something a real person can use and feel value from. That last mile — designing the experience, fitting it into reality — is where I want to keep working.

03Vision

Designing experiences people remember.

What I want to build is technology that doesn't stop at being a useful function — but lands in the world as something people return to, talk about, and feel a little changed by.

I've been drawn since childhood to the moment someone's emotion shifts. Composing world-building, narrative, technology, and operations into a single coherent experience has stayed at the core of what I do — in apps and AI, I care more about how naturally people slip into the world than how clever the system is underneath.

Theme parks are the ultimate form of experience design — space, story, technology, and operations all locking together to produce something memorable. That philosophy of integrating complex elements into a felt, lived experience is what I want to keep working toward.

04Projects

Projects

01 · 2021 →

AIbou

Native-level English learning app

Built in high school. A chat-based English partner for the situations Japanese learners actually face — booking a hospital appointment, ordering at a café. Connected to OpenAI before the public GPT-3 launch with a permission letter from the team. To gather a beta, I got a permit from the Shibuya police and spent weekends handing out posters around the station.

  • 1,800+ downloads
  • OpenAI integration (pre-public release)
  • Featured on TV
aibou.app
02 · 2025

AI Memory

Graph-theoretic memory for voice AI

Built during an internship — a memory architecture for an AI voice companion for older adults. A graph-theoretic 'water-flow' model decides not just what to recall, but how strongly — modeling gradual forgetting and the integration of episodic memory into semantic memory.

  • Patent pending
05Experience

Work Experience

06Honors

Honors

07Education

Education

08Media

Media

09Skills

Skills

Build

  • HTML · CSS · JavaScript
  • Next.js
  • Dart · Flutter
  • Swift
  • Python
  • Java

Design

  • Figma — UI/UX & Poster
  • Adobe Illustrator

Visual

  • Sony α7III — photography
  • Adobe Lightroom
  • DaVinci Resolve — film