Masamune Ishihara
Masamune Ishihara

Masamune Ishihara

Profile

I started programming at the age of 10 and have been creating mobile apps and web services as a hobby for 10 years by using React Native, Flutter, Rails, and Next.js. I got hooked on kaggle and am now a competition and notebook expert. Studying machine learning has offered me a lot of options in developing tools, so now I want to make web services and mobile apps by using this interesting technology.
I moved to Seattle from Japan last September (2020). I really love black tea, but since coming to Seattle, I’ve come to love coffee as there are many nice cafe options.
 

Bio

Shoreline Community College student
Live in Seattle
masaishi_masa*yahoo.co.jp (Please change * to at sign)
 

Education

 

Experience

Mar. 2023 Develop AI Bot Information Gathering Web App

I have created a web service that creates summaries and business ideas from papers and new Github repositories by Next.js.
While OpenAI's ChatGPT API does the work in summarizing and suggesting business ideas, I do the prompt engineering and also do the natural language processing preprocessing in python to improve the quality of the articles generated.
 
 

Jan. 2023 Develop Video Filter by Using Stable Diffusion

I used Stable Diffusion, which is essentially an image generation AI, to create a video filter.
Without ingenuity, the image would not be consistent with the one frame before and would re-generate a completely different image. I succeeded in improving the quality of the video by utilizing several image-processing techniques.
Animation filtered for cartoon-like effect
notion image
This is before after video.
notion image
 
 

Feb. 2023 Develop a browser game that can take any action depending on the text by using GPT-3.5.

I made a browser game with an AI Game Master powered by GPT, named TRPG_AI. You can do anything by typing actions like "ask for help" or "tame a dragon." The AI judges success rates.
 
 

Nov. 2022 Develop an image filter mobile app that converts images to illust style by Stable DIffusion.

I used textual inversion to make Stable Diffusion generate images that look like anime background illustrations, and I trained in multiple illustration styles. I then developed an image filter app by deploying my custom Stable Diffusion model to AWS Sagemaker and making it y-callable from a smartphone app built with Flutter.
↑ The notion video embedding did not work, so please double click on it and watch it in full screen🙏
 
 

Oct. 2022 Our team was in 2nd place at #HackSEA

Our team won 2nd place in #hackSEA, which is the hackathon by Hackathons International and Amazon
We were tasked with solving mental health issues for K-12 students.
During the hackathon, we came up with a solution that would allow students to complete activities to improve their mental health tasks as if they were a game, teachers could check their mental state with questions, and connect with friends who share their interests.
 
 

Oct. 2022 I received a silver medal for my entry in the “ボケ判定AIを作ろう!”(Let's make a joke judgment AI!) in Nishika.

I entered “ボケ判定AIを作ろう!”(Let's make a joke judgment AI!) and placed 19th at Nishika. Nishika provides data science competitions in Japan.
 
 

Apr. 2022 I received a silver medal for my entry in the “March Machine Learning Mania 2022 - Men’s” and become a kaggle competition expert.

I entered the March Machine Learning Mania 2022 - Men’s at Kaggle, and I received a silver medal for my efforts.
 
 

Apr. 2022 I become a kaggle notebook expert.

I got 5 bronze medals, so I become a kaggle notebook expert.
The below code is especially appealing. I join making a game bot competition and learn some reinforcement learning methods. Since none of the code other than the official code was published, we wrote and published the code to do q-learning.
 

Jan. 2022 I received a bronze medal for my entry in the “PetFinder.my - Pawpularity Contest”

I entered the PetFinder.my - Pawpularity Contest at Kaggle, and I received a bronze medal for my efforts.
The competition's purpose is to predict pets pictures’ popularity for rescue animals, which offers them a higher chance of finding loving homes.
I believed that pictures’ background is not important, so I used EfficientDet to detect animals. Then I used cropping pictures to train AI models.
 
I listed ideas in the discussion board below.
 
 

Dec. 2021 I completed the “How to Win a Data Science Competition: Learn from Top Kagglers” course

 
 

Oct. 2021 I developed a location analysis web service called “Tokyo Analyzer”

Opening a store in an urban city such as Tokyo has a very big risk because it is difficult to do business in a place where pedestrian access is poor. Thus, I developed location analysis AI which helps locate a good place to start a business; I hope it will help retail owners who want to open a new store and expand their business.
notion image
notion image
 
I trained AI for predicting the number of reviews on Google maps from around pedestrian distribution data.
notion image
 
I wrote an essay about this topic.
 
Presentation Slide
 
Product video
 
 

Nov. 2019 I worked on a team that developed a mobile app for Japanese learners called“Casual Japanese”

We received the Mobile CategoryBest Design Award at High School PC Concours 2019.
The team name was “devastatio”.
 
My team members and I developed a mobile app for learning Japanese.
 
 

Dec. 2014 I developed a flashcard web service called “Our Word Book”

I received an outstanding performance award at the 2014 International Ruby Programming Contest for Junior and Senior High School Students — In Mitaka.
I made a flashcard web service that user can make their own flashcard on web.
I used Sinatra to develop this site.
 
 

Dec. 2013 I developed a programmer assistance tool called “Programmer Secretary”

I received an outstanding performance award at the 2013 International Ruby Programming Contest for Junior and Senior High School Students — In Mitaka.
中高生国際Rubyプログラミングコンテスト
2013年12月7日、三鷹市の三鷹産業プラザにおいて「中高生国際Rubyプログラミングコンテスト2013 in Mitaka」の最終審査発表会が開催され、U-18部門、U-15部門の最優秀賞、優秀賞、奨励賞の各賞が決定しました。  最終審査に残ったのは、応募総数60件の中から、1次審査を突破したU-15の部5件、U-18の部4件の計9作品。各作品のプレゼンテーションの後、Rubyの生みの親であるまつもとゆきひろさんを審査委員長とする8名の審査委員によって最終審査が行われました。その結果、下記のとおり各賞が決定し、受賞者には表彰状と副賞としてノートパソコンやタブレットPCなどが送られました。  このコンテストの主催は、プログラミング言語「Ruby」の普及活動を行っている8団体で組織する実行委員会の他、三鷹市及び三鷹市教育委員会の主催により実施され、今回が3回目の開催となりました。  2014年度には、第4回を開催予定です。 11:30~ 開場/受付開始 12:00 開会 応募者によるプレゼンテーション 特別講演 講演タイトル:ロボット開発と国際親善ロボコン 登壇者:山北昌毅氏(東京工業大学大学院理工学研究科機械制御システム専攻 准教授) 〈講演概要〉 ロボット開発の場でどのようにプログラムが使われているのか、種々のロボットを 紹介しながらお話します。また、ロボットシステム開発に必要となる創造性を育む 国際親善ロボコンについて、今年ブラジルで開催された内容を中心に紹介します。 〈プロフィール〉 1962年三重県生まれ。1989年東京工業大学制御工学専攻修了(工学博士)。1989年同制御工学科大学助手、1993年豊橋技術科学大学情報工学系講師、1995年より東京工業大学制御システム工学科助教授。2000年より同大学大学院機械制御システム専攻助教授(2007年より准教授)。ロボット工学、適応学習制御、非線形ロバスト制御などの研究に従事。IEEE, 計測自動制御学会、日本ロボット学会などの会員。 審査結果発表・表彰式 入賞者には、賞状と副賞が贈られます。 審査委員長 まつもと ゆきひろ氏(「Ruby」開発者、一般財団法人Rubyアソシエーション 代表理事理事長) 審査委員 野田 哲夫氏(国立大学法人島根大学Ruby・OSSプロジェクトセンター長 教授) 田中 和明氏(国立大学法人九州工業大学大学院 情報工学研究院 機械情報工学研究系准教授) 笹田 耕一氏(Heroku, Inc.) 正村 勉氏(株式会社日立ソリューションズ サービスビジネス事業部長) 高橋 征義氏(一般社団法人日本Rubyの会 代表理事) 米田 昌弘氏(バイザー株式会社 代表取締役) 森 正弥氏(楽天株式会社執行役員 / 楽天技術研究所所長) 講評・講演 講演タイトル: Rubyが拓く未来 登壇者: まつもとゆきひろ氏 〈プロフィール〉 1993年よりプログラミング言語Rubyを開発している。1997年よりネットワーク応用通信研究所所属。2007年より楽天技術研究所フェローおよびRubyアソシエーション理事長就任。2011年よりHerokuチーフアーキテクト就任。2012年よりグルーブノーツチーフアーキテクト就任。 17:00 閉会 懇親会 ※上記スケジュールは変更となる可能性があります。あらかじめご了承ください。
中高生国際Rubyプログラミングコンテスト
I created “Programmer Secretary”, which is a tool to help programmers by creating templates and so on.