Publications

Writing published elsewhere, talks, papers, and awards.

2026

10.24Having Claude edit travel videos from Google Drive into a vlog with a homemade MCP (Japanese)ZennEXTERNAL 10.24Redesigning my personal blog with Claude Code, starting from a design system (Japanese)ZennEXTERNAL 09.30How MIXI built natural language search over billions of photos with Amazon S3 VectorsAWS Storage BlogEXTERNAL 07.27AI training, Day 1 (MIXI 2026 new-grad tech training)MIXI EngineersTALK 04.09A natural language search platform for hundreds of millions of photos and videos, and our road to adopting S3 VectorsAWS Connected CommunityTALK 03.11We made FamilyAlbum's huge media library searchable in natural language (Japanese)ZennEXTERNAL

2025

10Domain adaptation using non-parallel target domain corpus for self-supervised learning-based automatic speech recognitionSPEECH COMMUNICATION, Vol. 174, 103303PAPER 01MIXI AI new-grad training 2025MIXI EngineersTALK

2024

10.07From rules to ML: automatic photo print suggestions in FamilyAlbumShibuya Biz × AI: AI in business case study meetup #3TALK 09.04How FamilyAlbum's ML-based photo print suggestions work (Japanese)gihyo.jp (Gijutsu-Hyoron)EXTERNAL 09Hack Aichi 2024 (Aichi inter-university hackathon), Excellence AwardAWARD 03Domain adaptation of multilingual SSL-based speech recognition using only large-scale speech dataASJ Spring Meeting, 1-2-2PAPER 03Evaluating self-supervised speech recognition on speech by people with hearing impairmentsASJ Spring Meeting, 1-Q-23PAPER 03Improving speech recognition for older adults with an auxiliary age estimation taskASJ Spring Meeting, 1-2-5PAPER 03ASJ Spring Meeting, Student Excellent Presentation AwardAWARD 02Domain adaptation of multilingual SSL-based speech recognition using only a non-parallel corpusSPEASIP Workshop 2024PAPER 02Evaluating speaker adaptation for speech recognition of people with hearing impairmentsSPEASIP Workshop 2024PAPER 02Improving speech recognition for older adults through multi-task learning with age informationSPEASIP Workshop 2024PAPER 02Synthesizing an in-between voice of two speakers using the x-vector speaker spaceSPEASIP Workshop 2024PAPER —Customer testimonial for Amazon S3 VectorsAWS website (Amazon S3 Vectors)EXTERNAL

2023

10Domain adaptation with a non-parallel target domain corpusICAICTA2023PAPER 09Domain adaptation of SSL-based speech recognition models using only speech dataASJ Autumn Meeting, 2-Q-9PAPER 06Domain adaptation of self-supervised speech recognition models using target-domain speechSymposium on Music and Speech (Ongaku Symposium)PAPER 01.10Connecting learning and working: the career and work style I see after two months of work experience in my fourth year of university (Japanese)Shushoku Mirai Kenkyusho (Recruit)EXTERNAL

2022

09A response generation method of chat-bot system using input formatting and reference resolutionICAICTA-2022PAPER

2021

11A chit-chat response generation system using reformatted dialogue history19th Workshop on Informatics (WiNF2021), S-5-3PAPER