A small team. Each sees it through.

Joined Sony in 2006, working on embedded systems, device drivers and applications for mobile devices and surveillance cameras. Moved to Accenture in 2016, where he covered the full arc for manufacturing and public-sector clients — cloud platform builds, RFPs for system development engagements, estimation, project startup and delivery management. He went independent in June 2022 and founded TUWEI Innovation.

Overall management and business direction of the company. He also still works engagements as a technical lead. Recently he designed and built an AI knowledge search platform on GCP for a major power utility, and a Slack-integrated AI chatbot for a major telecom. Both use a RAG architecture built with LangChain, with answer quality tuned through chunk design, vector search and prompt design.

• Azure
• AWS
• GCP
• Architecture design
• CI/CD (Azure DevOps)
• Python
• TypeScript
• C#
• Java
• C/C++

• Microsoft Certified: Azure Solutions Architect Expert
• Azure AI Engineer Associate
• Azure Administrator Associate
• Azure Developer Associate
• DP-200 / DP-201
• DA-100 (Power BI)
• Fundamental Information Technology Engineer (Japan)

I started in embedded systems, then watched large engagements get delivered at a consulting firm. What struck me as the biggest waste was requirements being fixed without anyone knowing the actual operation — and then built exactly as fixed. Forward Deployed is our answer to that. Sit next to the client, see for yourself what the real problem is, and only then build. AI has become a strong enough tool. What is left is whether there is someone who can judge where to apply it.

After graduating from an information science university in China, he worked as a software engineer. He has built many web and interface systems on Azure and AWS PaaS, and has also worked on Azure-based IoT systems and data analytics platforms. Telecom, finance and IT infrastructure are his main fields. He served as PM on the project migrating Toyota Connected's back-office APIs to Azure. On a core system renewal for a public-sector affiliated organization, he joined from requirements definition and handled delivery management, leading a mixed onshore/offshore development team.

Defining and executing business and project strategy. He manages project scope, schedule, cost, quality and risk as one, and is accountable through to delivering work that meets the client's requirements. Coordination with clients, internal departments and external partners, team management, and reporting to executives are also his responsibility.

• Azure
• AWS
• LangChain
• Python
• C#
• Java
• React
• Kubernetes
• Docker
• CI/CD (Azure Pipelines)

• Microsoft Certified: Azure Solutions Architect Expert
• Azure AI Engineer Associate
• Azure Data Scientist Associate
• Azure Data Engineer Associate
• Azure IoT Developer Specialty
• Azure Developer / Administrator Associate
• MCSD: App Builder
• MCSA: Web Applications
• ITIL Foundation

In a large system renewal, what you decide during requirements definition stays with you to the very end. That is why I never move into design without understanding the client's operation. What I can say from leading teams that include offshore members is that good deliverables come from good questions. AI work is no different: before choosing a model, I start by confirming whether the operation is actually in a shape that AI can solve.

After graduating from Tokyo City University, he worked as a system architect, mainly in the finance and insurance industries, designing and implementing system architecture and cloud solutions on large projects. On a major bank's internet banking rebuild he designed and developed the frontend architecture (Angular); on an AWS rebuild of an insurer's system he designed and developed the backend architecture (Spring). By designing and building CI/CD pipelines he automated the path from development to deployment and shortened the release cycle. He joined TUWEI Innovation in April 2025.

Defining and carrying out the delivery strategy and execution plan for each project. He handles quality management and improvement, and manages teams including external vendors. For the RAG-based internal knowledge chatbot, he designed the search platform on Azure AI Search and developed the FastAPI backend running on Azure Functions.

• Azure
• AWS
• PyTorch
• TensorFlow
• Kubernetes
• Docker
• CI/CD (Jenkins, GitLab)
• React
• Angular
• TypeScript
• FastAPI
• Python
• Spring
• Java
• Gin
• Go
• Prometheus
• Grafana

• Microsoft Certified: Azure Solutions Architect Expert
• Azure AI Engineer Associate
• Azure Administrator Associate
• Azure Fundamentals

I have worked on systems in finance and insurance that are not allowed to stop. What that taught me is that building something that works is not enough. Can whoever inherits it fix it? When something breaks, can you trace it to the cause? That does not change for systems with AI in them. If anything, because accuracy fluctuates, monitoring and improvement have to be part of the design from the start.

After graduating from East China University of Science and Technology, he worked as a software engineer at HSBC and SAP SE, developing several large-scale IBM and SAP products on IBM AS/400, SAP R/3 and the SAP HANA Cloud Platform. On JERA's CX project he developed the frontend and backend of a system for analyzing power plant construction sites, running on Azure AKS and Function Apps. He owned database schema design, task management in Azure DevOps, code review, testing and deployment.

Design of SAP solutions and cloud application platforms, and development in the generative AI space. Alongside agent implementations using large language models, he works on image and video generation with Stable Diffusion, LoRA model training and inference API development.

• Azure
• Aliyun
• AWS
• PyTorch
• CUDA
• LLM agents
• Stable Diffusion
• Python
• PHP
• ABAP
• JavaScript
• HTML5
• SAP Fiori
• Swift
• Java
• Scrum

• Microsoft Certified: Azure AI Engineer Associate (AI-102)
• Azure AI Fundamentals (AI-900)
• Project Management Professional (PMP)
• ITIL v3 Foundation
• SAP Certified (BC100 / BC400 / HA100 / HA400 / SM100)

My work has ranged from core enterprise systems to generative AI. A heavyweight structure like SAP and a fast-moving technology like Stable Diffusion look like completely different things at first. But with both, the thinking is the same: where in the operation do you place it so that it actually helps. I try my hand at every new technology as it appears because I want to keep that sense sharp.

After completing the Graduate School of Applied Information Technology at The Kyoto College of Graduate Studies for Informatics, he worked as a systems engineer on Java-centered web application development, covering everything from the frontend (JSP/JavaScript) to the server side (Java). On claims and request intake systems for the life insurance industry he handled online completion of benefit claims and the intake platform for various policy procedures. On a fixed-asset management system he took on batch development centered on journal-entry customization, data migration, and fault investigation and repair. As a sub-leader he also managed progress and issues for offshore development and reviewed deliverables.

From defining project execution plans through development, quality management and quality improvement. For the internal RAG system he designed and developed the search platform using Azure AI Search and the FastAPI backend running on Azure Functions, and carried it through operational design and handover.

• AWS
• Azure
• Python
• Java
• JavaScript
• JSP
• HTML5
• Oracle
• MySQL
• PostgreSQL
• SQL Server
• R

• Microsoft Certified: Azure AI Engineer Associate
• Azure Developer Associate
• Azure Fundamentals
• DP-200
• TOEIC 760
• Bookkeeping Level 3 (JCCI)

I have built business systems in insurance and accounting, where mistakes are not allowed. What I learned then is that unless you understand the business rules behind the specification, you will not build something that behaves correctly. On the RAG system I work on now, improving search accuracy turned out to be, in the end, the work of understanding what the user wants to know. The technology changes, but the core of the job stays the same.