Naoki Shibuya, Developer in Tokyo, Japan
Naoki is available for hire
Hire Naoki

Naoki Shibuya

Verified Expert  in Engineering

Bio

Naoki is a senior machine learning engineer with experience in PyTorch. He is passionate about deep learning training, and he worked on model quantization and neural architecture search for vision models. Naoki is also an experienced C++ programmer who has worked on real-time algorithmic trading systems.

Portfolio

KiKaBeN
Python, PyTorch, Artificial Intelligence (AI), Deep Learning...
Woven by Toyota, Inc - Dojo
SQL, Data Visualization, Python, Data Science, R, Quarto, PyTorch, Pandas...
IOD Cloud Technologies Research Ltd.
Artificial Intelligence (AI), Machine Learning, Python, Deep Learning...

Experience

  • C++ - 20 years
  • Python - 12 years
  • Artificial Intelligence (AI) - 5 years
  • Technical Writing - 5 years
  • Deep Learning - 4 years
  • Computer Vision Algorithms - 3 years
  • PyTorch - 3 years
  • Quantum Computing - 1 year

Availability

Full-time

Preferred Environment

GitHub, PyTorch, Ubuntu, C++, Python

The most amazing...

...thing I’ve developed is a highly flexible (configurable) real-time message translation engine in C++. Many teams in the firm used it for their trading systems.

Work Experience

Solopreneur | Programmer | Researcher

2021 - PRESENT
KiKaBeN
  • Worked on a POC project for a paint shop scheduling problem, performing combinatorial optimization problems with nearly 900 cars. The objectives were color change minimization, model change maximization, and position change minimization.
  • Wrote technical blogs for machine learning (ML), deep learning, reinforcement learning, and mathematical concepts.
  • Worked on a POC project for a job shop scheduling problem and programming a hybrid solution combining an extensive neighborhood search (LNS) and QUBO-based optimization logic.
Technologies: Python, PyTorch, Artificial Intelligence (AI), Deep Learning, Deep Reinforcement Learning, Language Models, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Blogging, QUBO, Ising Models, Combinatorial Optimization

Data Analyst

2021 - 2023
Woven by Toyota, Inc - Dojo
  • Provided technical consulting for client companies. Researched and developed software using Python, Jupyter Notebook, R, RStudio, Shiny, and machine learning.
  • Handled data ingestion using API, data cleaning, and transformation to use clean data. Generated reports providing insights to senior management based on data from surveys and technical assessments using R, Quarto, Excel, and HTML/CSS.
  • Developed a dashboard UI using R Shiny and Plotly's Dash, using Python, HTML/CSS, and Plotly's graphical libraries.
  • Created technical assessments for Python, C++, machine learning, and computer vision.
Technologies: SQL, Data Visualization, Python, Data Science, R, Quarto, PyTorch, Pandas, Matplotlib, Plotly, RStudio, RStudio Shiny, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Deep Learning, APIs, Jupyter, Jupyter Notebook

Technical Writer | Consultant

2021 - 2022
IOD Cloud Technologies Research Ltd.
  • Wrote technical articles as per clients' requirements, adhering to their process and quality standards.
  • Participated in meetings to discuss and discover clients' needs for content creation.
  • Created and proposed draft structures for new AI-related articles.
Technologies: Artificial Intelligence (AI), Machine Learning, Python, Deep Learning, Technical Writing, Machine Learning Operations (MLOps), Content Writing, Writing & Editing

Senior Machine Learning Engineer

2020 - 2021
Edgecortix
  • Served as a senior machine learning engineer to optimize vision models for faster execution with post-training quantization.
  • Enhanced the deep neural network compiler using C++.
  • Wrote a neural architecture search (NAS) system to search for better architectures of vision models, using the ProxylessNAS paper.
Technologies: PyTorch, C++, TensorFlow, Artificial Intelligence (AI), AI Design, Object Detection, Classification, Computer Vision, Jupyter

Senior Robotics Software Engineer

2019 - 2020
Rapyuta Robotics
  • Developed a server-side program using Python and ROS messaging.
  • Conducted testing of our systems in customer sites with junior developers.
  • Handled communication in Japanese with clients for technical discussions.
Technologies: C++, Python, Robot Operating System (ROS), Robotics

Senior Research Engineer

2018 - 2019
Ascent Robotics
  • Acted as a research engineer in the company's decision-making team.
  • Built a prototype for driving scenarios using PyGame and Unity3D for testing reinforcement learning methods.
  • Presented in front of more than 100 people at the company's meetup, explaining the paper written by David Ha called "World Models."
  • Developed a demo using Mask R-CNN to detect the cars and pedestrians in the 3D simulation world (PyTorch.).
Technologies: PyTorch, Deep Reinforcement Learning, Artificial Intelligence (AI), Object Detection, Classification, Reinforcement Learning, Jupyter

Machine Learning Engineer

2017 - 2017
CLSA
  • Prototyped a system for predicting stock trading volume using a recurrent neural network.
  • Wrote a system to handle the daily trading data collection.
  • Served as one of the maintainers of the company's algorithmic trading engine.
Technologies: C++, C#, Python, Windows, Keras, Artificial Intelligence (AI), Code Review, Stock Market

Vice President

2009 - 2016
Deutsche Bank (Hong Kong)
  • Developed a new algorithmic trading system in C++ for the APAC markets and replaced the company's legacy system.
  • Made the new trading system quite flexible to enable faster delivery of new features and enhancements.
  • Interacted with salespeople, traders, and compliance officers for requirement gathering.
Technologies: C++, Python, Technical Leadership, Code Review, Stock Market

Vice President

2007 - 2009
Credit Suisse
  • Led the Asia-Pacific algorithmic trading IT team.
  • Enhanced and maintained the algorithmic trading engine together with my team.
  • Provided second-line support for sales and traders.
Technologies: C#, Microsoft Visual C++, Technical Leadership, Code Review, Stock Market

Assistant Vice President

2005 - 2007
Credit Suisse
  • Developed the company's liquidity provider system for their dark pool.
  • Developed the risk basket pricing system to handle quoting for institutional clients.
  • Provided the first-line support for proprietary traders while I sat among them.
Technologies: C#, Microsoft Visual C++, Technical Leadership, Code Review, Stock Market

Senior Developer

2004 - 2005
SIG Susquehanna
  • Developed the FIX-based broker gateway to handle client order flows.
  • Developed the Korean option fair volatility calculator.
  • Tested the company's exchange connectivity system for the Australian stock exchange.
Technologies: C#, Microsoft Visual C++, Code Review, Stock Market

Assistant Vice President

2001 - 2004
Lehman Brothers (Tokyo)
  • Managed both the exchange connectivity IT and the automated trading system teams.
  • Developed an XML-based message translation system for various communication protocols used by trading systems.
  • Developed the exchange connectivity systems and FIX connectivity systems.
Technologies: C++, SQL, Technical Leadership, Code Review, Stock Market

Senior Systems Specialist

2000 - 2000
Merrill Lynch (Tokyo)
  • Maintained the C++ program to upload trading data into the back-office database.
  • Maintained Perl scripts that invoke the C++ programs.
  • Supported the production system to ensure that batch processing worked without any issues.
Technologies: C++, SQL, Code Review, Stock Market

System Developer

1997 - 2000
Fujitsu Australia Software Technology (Sydney)
  • Designed Fujitsu's Enterprise JavaBeans container for their CORBA products.
  • Finetuned the JDBC driver for the Fujitsu Symfoware database for faster execution.
  • Traveled to Japan to support Fujitsu's engineers and customers.
Technologies: Java, C++, Code Review

Systems Analyst | Programmer

1996 - 1997
LINC International (Sydney)
  • Maintained and developed the company's multi-user travel booking system for Japanese travel agencies in Australia and Thailand.
  • Developed and maintained the database master using Microsoft Access as the front end to update records in Oracle for Windows NT.
  • Supported Japanese travel agents in Australia and Thailand.
Technologies: Visual Basic, Oracle

Upsampling with Transposed Convolution

https://naokishibuya.medium.com/up-sampling-with-transposed-convolution-9ae4f2df52d0
This is a blog article about understanding the upsampling with transposed convolution works for those who might have been confused about the logic. This is one of the posts I wrote for Medium, and it was the most popular one.

Behavior Cloning Driving Simulation

https://github.com/naokishibuya/car-behavioral-cloning
Built and trained a convolutional network for end-to-end driving in a simulator using Tensorflow and Keras. I implemented the driving (steering and throttle) model based on the Nvidia paper on deep learning and self-driving cars and using Udacity's car simulation environment for training and testing.

Link to the Nvidia paper used for this project:
https://devblogs.nvidia.com/parallelforall/deep-learning-self-driving-cars/

Deep Learning Examples for Medium Articles

https://github.com/naokishibuya/deep-learning
I've written different deep learning examples to accompany some of my Medium blog articles (http://naokishibuya.medium.com). I covered the topics such as style transfer, autoencoder, generative adversarial network (GAN), and transposed convolution.

Task Manager Project

https://github.com/naokishibuya/simply-done
This is the capstone project I did for Udacity's Cloud Developer Nanodegree in which I used AWS CDK (TypeScript) to allocate resources for AWS API Gateway, Lambda, S3, and DynamoDB and use them for the simple to-do list manager. I also used React, Redux, and Redux Thunk for the front end.

Transformer's Encoder-Decoder Architecture

https://naokishibuya.medium.com/transformers-encoder-decoder-434603d19e1
I've written a series of articles on transformer architecture:
• Encoder-decoder architecture
• Self-attention mechanism
• Positional encoding
• Word embedding
• Beam search
• BLEU
I've also drawn technical diagrams to explain the details of the architecture.
2003 - 2008

Master's Degree in Applied Finance

Macquarie University - Sydney, Australia

1988 - 1992

Bachelor's Degree in Physics

Shizuoka University - Shizuoka, Japan

MARCH 2023 - MARCH 2026

AWS Certified SysOps Administrator - Associate

Amazon Web Services

JUNE 2021 - PRESENT

Cloud Developer Nanodegree

Udacity

MARCH 2021 - MARCH 2026

AWS Certified Cloud Practitioner

Amazon Web Services, Inc.

FEBRUARY 2021 - DECEMBER 2026

AWS Certified Machine Learning – Specialty

Amazon Web Services

JANUARY 2021 - PRESENT

Computer Vision Nanodegree

Udacity, Inc.

JANUARY 2019 - PRESENT

Deep Reinforcement Learning Nanodegree

Udacity, Inc.

JULY 2018 - PRESENT

Robotics Software Engineer Nanodegree

Udacity, Inc.

FEBRUARY 2018 - PRESENT

Deep Learning Specialization

DeepLearning.AI

OCTOBER 2017 - PRESENT

Self-driving Car Engineer Nanodegree

Udacity, Inc.

JULY 2016 - PRESENT

Data Science Retreat Bootcamp

Berlin, Germany

JULY 2016 - PRESENT

Machine Learning

Stanford University | via Coursera

MARCH 2016 - PRESENT

Machine Learning Engineer Nanodegree

Udacity, Inc.

Libraries/APIs

PyTorch, NumPy, TensorFlow, Matplotlib, Keras, Scikit-learn, React, Pandas

Tools

Slack, Jupyter, GitHub, Amazon SageMaker, Microsoft Visual C++, Redux Thunk, Plotly

Languages

C++, Python, C#.NET, C#, R, SQL, Java, Visual Basic, Octave, TypeScript

Frameworks

Apache TVM, RStudio Shiny, Redux

Platforms

Linux, MacOS, Oracle, Windows, Amazon Web Services (AWS), Ubuntu, AWS Lambda, RStudio, AWS Elastic Beanstalk, Jupyter Notebook

Storage

Amazon S3 (AWS S3), Amazon DynamoDB

Other

Deep Learning, Convolutional Neural Networks (CNNs), Machine Learning, Computer Vision, Computer Vision Algorithms, Artificial Intelligence (AI), Technical Leadership, Code Review, Deep Reinforcement Learning, Technical Writing, Blogging, Stock Market, Robot Operating System (ROS), Quantitative Finance, Simultaneous Localization & Mapping (SLAM), AWS Certified Cloud Practitioner, Amazon Machine Learning, AWS Certified Solution Architect, Amazon API Gateway, AI Design, Object Detection, Classification, Reinforcement Learning, Natural Language Processing (NLP), Technical Drawing, Robotics, APIs, Financial Modeling, Data Science, Statistics, Dash, Machine Learning Operations (MLOps), Language Models, Generative Pre-trained Transformers (GPT), Data Visualization, Quarto, Content Writing, Writing & Editing, QUBO, Ising Models, Combinatorial Optimization, Quantum Computing

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring