Software and Infrastructure developer with 9+ years of experience, currently leading engineering teams with a focus on generative AI and AI agent development. AWS Certified Solutions Architect - Professional with deep expertise in serverless architecture and production-level AI systems. Published 3 technical books in 2025 on MCP and AI agents. Led development of A-BOSS, a production AI agent serving 700+ users at KDDI. Camera and music-loving web developer who is also an amateur DTM composer and guitarist. Active open source contributor with 190+ GitHub repositories.
Leading generative AI application development with focus on AI agents. Development lead for A-BOSS (AI Business Department Manager) serving 700+ users. Published 3 technical books in 2025. Contributing to AWS Japan Official Blog and providing training as KAGAI Learning Hub instructor.
Development lead for A-BOSS - AI agent for KDDI corporate sales, serving 700+ users with 10 hours/month efficiency gain per user
Built 議事録パックン (Minutes Pakkun) - comprehensive sales support tool with meeting transcription and proposal generation
Deployed かぐたん (Kagutan) - secure internal AI chatbot to 1,200 users across 4 companies with ~2,000 monthly requests
Co-authored 3 technical books: 'やさしいMCP入門', 'AIエージェント開発/運用入門', and 'KAG Tech Book Vol.1'
Featured in AWS Japan Official Blog for A-BOSS case study
Instructor at KAGAI Learning Hub for GenAI and AI agent training
Currently on parental leave (Sep 2025 - Apr 2026) while continuing technical activities
As an SRE Engineer, improved continuous development processes using AWS CDK, GitHub Actions, and Autify.
Built and improved CI/CD pipeline and infrastructure on AWS
Migrated services from EC2 to ECS/Fargate with Blue/Green deployment
For the Michibiku application, we conducted an AWS Foundational Technical Review (FTR) and obtained AWS Certified Software.
Expanded automated testing using Autify and developed execution pipeline for QA efficiency
Full-cycle engineer for Web-based event matching platform using AWS, Nest.js, and React with TypeScript. Development Manager for Product Infrastructure team.
Developed pseudo-live function for pre-recorded video live streaming
Built video distribution platform and API infrastructure for external collaboration
Developed public API platform and MA/SFA integration (Marketo/Salesforce connectors)
Created Data Clean Room with Snowflake and AWS
Led Product Infrastructure team mission formulation and execution
Accelerated CI/CD and maintained E2E tests for engineering team support
Introduced FaaS migration and Blue/Green deployment
Tech Lead and AWS architect for Web-based and Native App chatbot services. Led team of 8-10 members with responsibilities spanning architecture, development, team management, and stakeholder negotiation.
First experience as tech lead
Architecture featured at AWS Summit Tokyo 2019
Managed 8-10 member team
Scrum development methodology
Developed data analysis infrastructure for chatbot conversation analysis. Implemented Data-lake architecture with serverless Lambda Python and Pandas.
Data-lake Architecture using AWS
Serverless with Lambda Python
Pandas data analysis
Scrum development
Developed and operated Web-based chatbot services using AWS, Python, and Node.js. Architecture featured ECS with Docker, DynamoDB, Lambda, and S3.
First experience using AWS
First experience using Python and Node.js
Serverless architecture implementation
CI/CD with Drone and Pytest
Lectured at AWS Summit Tokyo 2019 company booth
Scrum development
Led analytics operation system development using Splunk and Hadoop (MapR) for internal log analysis.
Project management and team building
Big data handling with Splunk and Hadoop (MapR)
Internal log analytics operation system
Developed and operated location-based services using on-premise servers. Built integrated monitoring system with Zabbix.
Location-based service development and operation
Integrated monitoring system with Zabbix and Expect automation
First professional engineering role
Serving 700+ KDDI corporate sales staff in production
Achieves 10 hours/month efficiency gain per user
Reduced proposal creation time by up to 1 hour
Contributes to KDDI-BX initiative (46,939 hours/year saved company-wide)
Featured in AWS Japan Official Blog and multiple media outlets (Nikkei, Business Insider Japan)
Multi-modal analysis including PowerPoint documents
Automatic user feedback summarization and reflection
RAG with internal knowledge base
LLMOps with Langfuse for observability
Saves up to 1 hour per meeting
End-to-end sales activity support
Multi-modal input (audio/video)
Automated proposal generation
Internal knowledge base integration
Competitor analysis automation
16 stars on GitHub
Early contribution to MCP ecosystem
TypeScript implementation
Integration with midi-writer-js
Production-ready Terraform module
Cost-optimized AWS deployment
Comprehensive blog posts on operation
Community contributions to Langfuse ecosystem
Hybrid architecture with chatbot and human operators
Elasticsearch-based FAQ search engine
Labor saving and improved customer satisfaction
Roles: Primary Developer, Data Scientist, Scrum Master
Native app chatbot for Android and iOS
WebSocket-based real-time communication
Microarchitecture with Docker containers
Featured at AWS Summit Tokyo 2019
Active participant in Japanese Langfuse community, sharing LLMOps knowledge and best practices.
Speaker at Langfuse Night #1 (Nov 2024): 'Langfuse v3 使ってますか?'
Developed langfuse-v3-terraform module for AWS deployment
Published blog posts on cost-effective Langfuse operation
Maintaining 190+ public repositories with focus on developer productivity tools and AI agent ecosystem.
Created midi-mcp-server (16 stars) - MCP server for MIDI generation
Developed mcp-prompts - MCP server for prompt template management
Built langfuse-v3-terraform - Terraform module for Langfuse v3 on AWS
Maintained portfolio and blog sites with 100% Lighthouse scores
Created developer tools: ebook_homebrew, update_release, npm_profile_card
A-BOSS project contributed to 46,939 hours/year saved across KDDI-BX initiative. Generated 10 hours/month efficiency per user for 700+ sales staff.
Stable and flexible system design utilizing AWS and successfully responded to large-scale traffic without problems.
Co-authored with Minoru Onda. Introduction to Model Context Protocol (MCP), the new standard for AI agent era. Based on popular slides, explains fundamentals accessible to both engineers and non-engineers. Covers MCP architecture, hands-on tutorials, client/server applications, and business impact. Reprinted within 2 months due to strong reception. 160 pages, A5 format.
Co-authored with Minoru Onda and Masaki Tsukada. Comprehensive guide covering AI agent fundamentals to implementation and LLMOps. Includes complete implementation examples from simple reasoning agents to MCP-based and multi-agent systems. Covers LangChain, LangGraph, Mastra, Amazon Bedrock, Langfuse, and Guardrails. Features full-color illustrations and extensive hands-on exercises. Launch event held on release date. 408 pages.
Technical anthology by KDDI Agile Development Center engineers. Contributed chapter on 'Celery in Detail' - comprehensive guide to Python distributed task queue library including operational considerations and AWS scaling practices. Published at Gijutsushoten 18 (Tech Book Festival 18).