Naoki Takano's Email Addresses & Phone Numbers

📖 Summary

Voice Actor @ Twitter 広告システムのための日本語訳のナレーションを私の声で行いました。英語しか話せない同僚とのやり取りは困難を極めましたが、日本の方々のツイッター広告システムの理解に大きく貢献したと思います。 From March 2013 to March 2015 (2 years 1 month) San FranciscoPlaying Manager @ Twitter loving hands on From August 2012 to March 2015 (2 years 8 months) Software Engineer @ Convoy Inc Software Engineer @ MODE, Inc. Design and implemented scalable and high available backend system for IoT using Golang.Design and implemented IoT SDK on various platform(iOS, Android, Broadcom, TI and embedded Linux)Design and implemented management console UI for IoT. From March 2015 to May 2019 (4 years 3 months) San MateSoftware Engineer @ Twitter ‮Implemented RTL support on twitter.comPlaying manger. Guardian and God of Twitter celebrities. From September 2011 to March 2015 (3 years 7 months) San Francisco Bay AreaSenior Software Engineer @ PayPal, USA Developed secured payment feature for high risk gambling merchant in Europe and Australia. According to the merchant risk and credit card information, the feature required credit secure code. Needed to investigate legacy and complicated code base to implement this feature on Linux platform.Support live issue fix team and analyzed and fixed critical issues quickly. I fixed merchant related transaction issues and eBay transaction issues From June 2010 to September 2011 (1 year 4 months) Senior Software Engineer @ Ziosoft Developed algorithm development platform for medical image processing. The platform made the algorithm developer easier to develop and maintain rather than simple C/C++ language. The platform provides special language, debugger and test environment to process multiple data in multiple computers concurrently.Optimized registration algorithm to more than one hundred times faster than original. The optimization uses SIMD instruction (i.e. SSE, SSE2, SSE3, SSE4.1) and an algorithmic approach. Reduced the algorithm big O from O(n^2) to O(n) and more.Optimized 3D volume filter algorithm in CUDA environment. CUDA is GPGPU development platform from nVidia and faster than CPU only computation in some high performance computing area. I used both CPU and GPU according to each algorithm efficiently. Developed MR Cardiac application. This application automatically detects the heart component in MR data. Motion analysis is done with a statistical method. This application dramatically reduces the diagnostic time of medical doctors. Also developed the GUI. Implemented segmentation algorithm for medical CT images of the human head and heart. This program automatically removes the bone component in human head or heart data, and displays only the blood vessel components or other organ parts. This algorithm shows the shape of the vessels and makes medical doctors easily detect aneurism, calcification and other defects in the human head or heart. Set up network environment (Apache, Subversion, and DHCP server, etc.), tools, and Linux and Windows machines for Ziosoft's start-up in the U.S. Because I was the only one engineer when we started Ziosoft in U.S. and there was no IT department. Also provided communication support between the Japan team and the U.S. team. From January 2006 to May 2010 (4 years 5 months) Software Engineer @ Ziosoft, Inc Optimized volume rendering engine to make it faster. The method used intrinsics for SIMD (SSE, SSE2, MMX) from normal C++/C language, but also, from an algorithmic point of view, changed the design and implementation. And used multi-threading efficiently to use all CPU power. Designed and implemented real-time clustering platform for volume rendering. It took a long time to process volume data to render images. The time depended on the data size and quality. To get a high resolution image took several seconds. We used a real-time volume clustering engine to reduce the time. The system supports up to eight server machines.Designed and implemented Web-based medical image viewer. Users can view CT, MRI, and other medical images remotely, not only 2D images but also 3D volume rendering images. Because the 3D volume rendering task is very heavy, an image processing server provides the rendering service and then multiple Web-based viewers display the result through the network. Used Active-X, COM, DCOM, TCP/IP. Designed and implemented remote function call framework. Because volume processing uses so much memory and CPU power, separated into two parts, one to display client part and one to process volume data part. Using this remote function call framework, we could make client part much lighter. After this improvement, we could use small machines like notebook-PCs as thin-client viewers. Developed manual edit tools for volume processing. In volume rendering, users want to see only a specific part. These tools manually erase arbitrary parts of the volume data. For example, they cut cross-sections and erase continuous areas. Used MFC library.Supported Web-based search system for medical data listing.This system uses ASP, COM, and Javascript. And developed Active-X component to provide API for third vendors to use the functions in our products. From September 2001 to January 2006 (4 years 5 months)