About

avatar

Minato Sato

Lead Software Engineer (Embedded AI)
Sony
githubGitHublinkedinLinkedinxX

About Me

For more than nine years, I’ve been building AI-powered products that bridge human creativity and machine intelligence. My work spans embedded AI, computer vision, generative AI, recommendation systems, data science, and technical leadership—always with a focus on turning research and ideas into reliable products.

My journey into AI began 15 years ago while building robots at university. Facing the limits of traditional control systems, I realized that what machines lacked was not motion, but understanding. That realization has guided my path ever since.

At Sony, I’m a Lead Software Engineer in embedded AI, leading a development team that builds subject-recognition and tracking features for mirrorless cameras. Our work includes people tracking and AI-powered autofocus, helping photographers concentrate on expression and timing instead of focus mechanics.

Technical excellence matters, but it is not where direction begins. I ground product and technical decisions in what customers tell us and what photographers experience in the field, then translate those insights into priorities for the team.

I’m a hands-on technical lead rather than a full-time manager. I work directly on the hardest engineering problems while setting technical direction, reviewing architecture and design, and making the final calls on quality, performance, and resource trade-offs. I also build the technical foundations and development practices that help the team reduce cognitive load, rework, and technical debt while continuously delivering maintainable software. My responsibility is not only to write strong code, but to enable the team to deliver the right product with dependable quality over the long term.

Until February 2026, I also worked as an independent AI consultant and research engineer, partnering with organizations from product strategy and prototyping through production. My work included:

  • LLM applications, RAG, GraphRAG, and AI agents
  • Language, speech, and real-time conversational AI
  • Recommendation systems powered by embedding-based preference modeling
  • Local LLM and inference environment evaluation

My approach is grounded in pragmatic engineering and continuous learning: prototype quickly, learn from real users, and refine systems until advanced technology becomes dependable and useful. I enjoy working across the stack—from algorithms and low-level optimization to product design and technical direction.

Looking ahead, I want to create technology that amplifies human intelligence and creativity, helping people see, think, and make in ways that once felt beyond reach. I’m also interested in how autonomous, human-centered organizations can unlock the creativity of teams.

On this website, I write primarily in Japanese about natural language processing, machine learning, and software engineering.

GitHub / Twitter / Facebook / LinkedIn / Google Scholar

Skills

  • AI / ML: Embedded AI, computer vision, object tracking, LLMs, RAG, GraphRAG, AI agents, recommendation systems, speech and language processing
  • Engineering: C++, C, Python, Cython, TypeScript, RTOS, embedded systems, hardware-aware optimization
  • Data / Cloud: MySQL, Impala, AWS, GCP
  • Leadership: Technical direction, architecture and design review, product quality, team development, technical consulting

Work Experience

Lead Software Engineer (Embedded AI)

Sony Corporation
Yokohama, Kanagawa, Japan
July 2026 – Present

I lead a development team building embedded subject-recognition and tracking AI for mirrorless cameras, including people tracking and AI-powered autofocus.

As a hands-on technical lead, I contribute directly to implementation while setting technical direction, reviewing architecture and design, and making final decisions on quality, performance, and constrained hardware resources. I set these priorities based on customer feedback and real-world user needs, rather than technology alone.

I also support planning and task decomposition, mentor engineers through technical decisions, and improve how the team develops software. A key focus of my current role is building a sustainable development system—one that enables the team to continuously deliver maintainable software while systematically reducing technical debt. I’m accountable for the technical outcome and product quality delivered by the team.

Senior Software Engineer (Embedded AI)

Sony Corporation
Yokohama, Kanagawa, Japan
April 2022 – June 2026

Developed and optimized embedded subject-recognition and tracking AI for mirrorless cameras, including people tracking and AI-powered autofocus. My work spanned algorithms, low-level software, hardware integration, performance optimization, field testing, and product refinement.

Products contributed to: α7RV, α9III, α1II

AI Consultant / Research Engineer

Freelancer
Yokohama, Kanagawa, Japan
February 2023 – February 2026

I partnered with companies to design and deploy AI systems spanning LLM applications, RAG and GraphRAG, AI agents, recommendation engines, language and speech processing, and local inference environments.

My role bridged strategy and execution: translating research into production and ensuring that advanced technology became practical, scalable, and meaningful.

Each engagement was driven by agile co-creation—rapid prototyping, feedback-driven iteration, and a commitment to delivering measurable impact early.

Machine Learning Engineer / Data Scientist

So-net Media Networks Co., Ltd.
Shinagawa, Tokyo, Japan
October 2018 – June 2022

Led the end-to-end development of AI-powered user targeting systems for digital advertising.
By combining large-scale data pipelines with adaptive machine learning models, I improved targeting precision and contributed to significant revenue growth.
Focused on building systems that balanced innovation with operational reliability.

Product: VALIS Engine

Software Engineer

Sony Imaging Products & Solutions Inc.
Atsugi, Kanagawa, Japan
April 2017 – September 2018

Developed an “AV over IP” solution for Sony’s CLED—an ultra-high-resolution LED display technology.
Collaborated across hardware and software teams to create efficient, low-latency video transmission frameworks for large-scale visual systems.

Data Scientist / Analyst (Part-time)

National Maritime Research Institute (NMRI)
Mitaka, Tokyo, Japan
November 2014 – March 2017

Analyzed ship operation monitoring data to evaluate propulsion performance and fuel efficiency.
Published multiple research papers proposing new methods for modeling and interpreting maritime data.

Education

The University of Electro-Communications, Tokyo, Japan (2015–2017)

  • M.Eng. in Social Intelligence and Informatics
    • Salutatorian (2nd in department)
    • Research: Natural Language Processing using Deep Learning

The University of Electro-Communications, Tokyo, Japan (2011–2015)

  • B.Eng. in Informatics
    • Research: Evolutionary Machine Learning (Reinforcement Learning)
    • Developed ultra-small exploration rovers and nanosatellite systems

Publications

Google Scholar Profile

Conference Papers (Reviewed)

  • Minato Sato, Ryohei Orihara, Yuichi Sei, Yasuyuki Tahara, Akihiko Ohsuga: Text Classification and Transfer Learning Based on Character-Level Deep Convolutional Neural Networks, in Agents and Artificial Intelligence, Springer, pp.62–81, 2018.
  • Minato Sato, Ryohei Orihara, Yuichi Sei, Yasuyuki Tahara, Akihiko Ohsuga: Japanese Text Classification by Character-Level Deep ConvNets and Transfer Learning, ICAART 2017, Vol. 2, pp.175–184, 2017 (acceptance rate: 26.5%, Best Student Paper Award nominee).
  • Minato Sato, Kotaro Usui, Masaya Nakata, Keiki Takadama: Detecting Shoplifting From Customer Behavior Data by Extended XCS-SL, IEEE CEC 2015, Sendai, Japan.

Regional Papers (Reviewed)

  • 佐藤挙斗ほか: 「文字レベル深層学習による日本語テキストの分類と転移学習」, JAWS2016, 2016/09.

Regional Papers (Unreviewed)

  • 佐藤挙斗ほか: 「文字レベル深層学習によるテキスト分類と転移学習」, 人工知能学会 SIG-FPAI, 2016/12.
  • 佐藤挙斗ほか: 「文字レベル深層学習の日本語データセットへの応用」, 情報処理学会 SIG-ICS, 2016/08.
  • 加納敏幸, 佐藤圭二, 佐藤挙斗: 「運航モニタリングデータによる船舶推進性能評価」, 日本船舶海洋工学会秋季講演会, 2015/11.
  • 小峯嵩裕, 佐藤挙斗ほか: 「動的環境におけるアーカイヴ型学習分類子システム」, 進化計算シンポジウム, 2013/12.

Achievements