ghost & ink
Released5 AI personas simultaneously destroy your idea, then rebuild something unexpected from the wreckage. Built with 5 parallel SSE streams on Cloudflare Workers + Hono. Supports a free demo mode (Workers AI) and BYOK mode (OpenAI).
Hi, I'm
Eiichiro Iriguchi.
Backend Engineer — Azure × LLM
Building systems that deliver excitement to people.
scroll
who i am
About
Backend engineer with expertise in both Azure cloud infrastructure and LLM application development. I take full ownership from architecture design to production release.
I excel at identifying problems and transforming entire workflows — for example, I introduced an automated testing foundation to a team that had zero tests and relied on manual Excel sheets, reducing manual QA overhead by 96%. I have hands-on experience with Azure OpenAI Service, RAG, and AI agent architectures, and can handle the entire lifecycle of LLM applications in production.
Everything I build — engineering or creative — stems from one core motivation: to deliver excitement to people.
technologies
Skills
Work Personal Learning
Cloud / Azure
Azure Functions Azure Blob Storage Azure Entra ID Azure OpenAI Service Azure CI/CD
AI / LLM
LLM Application Dev Prompt Engineering RAG Architecture AI Agent Systems
Backend
C# ASP.NET Core Python FastAPI Node.js Hono REST API Design SQL Server
Frontend
TypeScript Angular React Next.js
Infrastructure
Docker GitHub Actions Vercel Cloudflare
Tools
Git / GitHub VSCode Linux
selected work
Projects
DEBUG-ZERO
WIPAn open-source card-based board game. Designing the game mechanics, lore, and rules from scratch. Building a real-time multiplayer foundation with Hono, React, and Cloudflare Workers (Durable Objects).
Hono React Cloudflare Workers Durable Objects Open Source
AI Bot Management Platform
ReleasedBuilt to solve a bottleneck where knowledge base registration required developer involvement. Designed a non-engineer-friendly management UI and shipped solo in one month. Enabled non-developers to register and manage knowledge bases directly, improving operational efficiency. Implements multi-bot configuration, LLM conversation settings, and KB lifecycle management.
Azure OpenAI ASP.NET Core TypeScript Azure Blob Storage
RAG Portfolio API
ReleasedBuilt to demonstrate RAG implementation skills directly — using myself as the use case. Recruiters can ask questions about my skill sheet in natural language. In production I run the same architecture on Azure AI Search + Azure OpenAI. This portfolio version uses free-tier alternatives (Qdrant Cloud, Groq + Llama 3.3) — same design, zero cost. Built with FastAPI, LangChain (LCEL), and SSE streaming.
RAG FastAPI LangChain Qdrant Python
RAG Evaluation Platform
ReleasedBuilt to solve a common problem: RAG quality is often judged by feel. Automatically scores output quality (Relevance, Faithfulness, Completeness) and visualizes it in a Next.js dashboard. Integrates with 3 lines of code. Supports self-hosting.
RAG Evaluation FastAPI Next.js Supabase Python
Obsidian × Claude Code Workflow
ActiveAutomated daily report generation and knowledge management workflow using Claude Code and Obsidian Vault.
Claude Code Obsidian Automation TypeScript
zenn articles
Blog
get in touch
Contact
Work inquiries, technical discussions, or just a chat — all welcome.