Profile

Hi, I'm Lukas from Japan. I have a background in Machine Learning and AWS. Currently, I'm doing full-stack development. You can see my projects here . As of recent, I'm into Neovim and competitive programming.

LinkedIn

GitHub

Skills

  • Natural Languages

    Flag Icon

    Japanese - Native Proficiency

    Flag Icon

    English - Native Proficiency

    Flag Icon

    German - Native Proficiency

  • Programming Languages

      SQL

  • Frameworks / Tools

Work Experience

Cloud Support Engineer

Amazon Web Services Japan

2022/04 - 2023/06

  • Provided corporate customers with technical assistance pertaining to Big Data and Machine Learning services

  • Work entails

    • Troubleshooting production scale issues

    • Assisting clients engineer new projects

    • Collaborating internally with an international team

  • Extensive usage and reasearch of related OSS

  • AWS Services handled include

    • SageMaker (Machine Learning)

    • Glue (ETL / Apache Spark)

    • Athena (SQL Analytics)

    • DynamoDB (NoSQL Database)

ML Research Intern

Sony Group Corporation

2022/02 - 2022/03

  • Joined the Computer Vision team at the R&D Department for 1 month

  • Conducted research on Object Tracking methods that utilize deep learning, focusing on improving accuracy for lightweight models

Education

M.S. Computer Science

Osaka University

2020/04 - 2022/03

  • Computer Science and System Engineering major with a focus on Computer Vision

  • Conducted research on lightweight Object Detection for embedded systems with a focus on energy consumption and processing speed.

  • Utilized Python, Docker, Jupyter, Pytorch, Tensorflow for a total of 3 years

  • Experience with accelerators / boards such as Google Edge TPU / Nvidia Jetson Nano

B.S. Computer Science

Osaka University

2016/04 - 2020/03

  • Computer science major with a focus on Software Engineering

  • Extracurricular activity: English Debate Club

Publications & Awards

Journal article

Lukas Nakamura, Hiromitsu Awano, "Temporal Ensemble SSDLite: Exploiting Temporal Correlation in Video for Accurate Object Detection", IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences E105.A(7), 1082-1090, 2022

2022/07/01

  • Improving accuracy up to 4.9% while keeping inference speed for Video Object Detection using deep learning methods by leveraging ensembles (combining multiple models)

  • Link

Excellence Award - Hackathon

Yahoo Hackathon - Open Hack U 2021 Online Vol.3

2022/02/26

  • Created a robot companion that changes interaction style depending on the situtation at home

  • Utilized computer vision for scene recognition in a custom environment by creating a custom dataset

  • Summary page of this Hackathon

Finalist - English Debate Tournament

All Japan Freshman Tournament - Momiji Cup 2016 @ Kyoto Univ.

2016/10/15-16

  • Finalist (top 2 out of ~100 teams)

  • Teams from universities all over Japan

  • Other awards include 2x Semi-finalist in all Japan open tournaments (free for anyone to participate)