takuti.me

I'm working on machine learning and data science at Arm Treasure Data, an Arm's internal organization developing large-scale enterprise customer data platform. My mission is to connect theory and practice in their B2B business. So far, I have been practically acted as a data scientist, data analyst, consultant, machine learning engineer and software engineer through the implementation of scalable ML library, out-of-the-box ML application, and customer-specific solutions on the wide variety of data.

home Nakano, Tokyo, Japan
mail k.takuti at gmail.com

twitter twitter.com/takuti
github github.com/takuti
images speakerdeck.com/takuti
linkedin linkedin.com/in/takuti
instagram instagram.com/takuti
tumblr takuti.tumblr.com
rss takuti.hatenablog.com

Things I like

Traveling. In Japan, I like countryside and small cities like my hometown, Nagano. Oslo (Norway) and Vancouver (Canada) are two of the best foreign cities I've ever visited.
Craft beer, especially Amber Ales, and Japanese sake. While traveling, I enjoy visiting breweries and drinking something local.
Burgers. Good beer and good burger, that's the combination.
Recommender systems. In this field, academic theory and industrial practice clearly intersect one another. My recent interest is in UI/UX-related consideration from the HCI perspective.
Matrix computations. Singular Value Decomposition is particularly awesome in terms of simplicity, applicability, and extensibility.
Walking, running, hiking. Those are the way to spend my holiday.
Writing. I have published a bunch of blog articles as well as some research papers.
Presentations. Talking in front of many audiences is always exciting and stimulating. I'm continuously seeking such opportunities.

Recent blog articles

» more

Conference talks and publications

Query-Based Simple and Scalable Recommender Systems with Apache Hivemall @ RecSys 2018
filePaper / imageSlides / videoVideo
Apache Hivemall: Query-Based Handy, Scalable Machine Learning on Hive @ ODSC Europe 2018
imageSlides
FluRS: A Library for Streaming Recommendation Algorithms @ EuroSciPy 2017
videoVideo / imageSlides / codeCode
Sketching Dynamic User-Item Interactions for Online Item Recommendation @ CHIIR 2017
filePaper / imagePoster / codeCode
Incremental Factorization Machines for Persistently Cold-starting Online Item Recommendation @ RecProfile 2016 (in conjunction with RecSys 2016)
filePaper / imageSlides / codeCode
User Modeling in Folksonomies: Relational Clustering and Tag Weighting @ WIMS 2015
filePaper / imageSlides / codeCode