I build grounded AI applications with Java, Spring Boot, LLMs, RAG, and Agent workflows.
Currently building JianMiao 简喵, a job-description-grounded AI career preparation platform that connects resume scoring, job description diagnosis, evidence-grounded rewriting, Guard Trail, interview preparation, cover letter generation, and job-session workbench into one auditable workflow.
- Java / Spring Boot backend engineering
- LLM application engineering
- RAG and evidence-grounded generation
- Agent workflow design
- Structured output validation
- Anti-hallucination guardrails
- AI product architecture
- Career-tech product development
A job-description-grounded AI career preparation platform.
JianMiao is not a generic AI resume builder. It is designed around a more specific workflow:
saved resume version
→ target job description
→ job description diagnosis
→ candidate evidence indexing
→ evidence-grounded resume rewriting
→ Guard Trail
→ interview preparation
→ cover letter
→ job-specific material package
The core goal is to help users create job-specific application materials that are more factual, traceable, and interview-defensible.
Instead of only making resume text look better, JianMiao focuses on a stricter question:
Which parts of the AI-generated content are actually supported by the candidate's existing experience?
Key modules:
- Resume scoring
- Job description diagnosis
- Candidate evidence indexing
- Job-specific resume rewriting
- Guard Trail for unsupported or risky claims
- Evidence-constrained bullet rewriting
- Grounded Interview Pack
- Evidence-constrained cover letter generation
- JobOptimizationSession Workbench
A Spring AI based multilingual customer-service Agent platform for cross-border e-commerce scenarios.
Focus areas:
- Java / Spring Boot backend
- LLM-powered customer-service workflow
- Multilingual support
- Agent-style task orchestration
- AI application engineering
I am interested in the engineering layer between raw LLM capability and real user-facing products.
My current direction includes:
- grounded generation
- evidence chains
- claim verification
- structured output validation
- anti-hallucination design
- workflow-level AI products
- reliable AI systems for real users
I write about AI application development, Java backend engineering, LLM products, RAG, Agent workflows, product architecture, and the process of building JianMiao.
- Blog: https://yanxai.com
- GitHub: https://github.com/RyanCoreAI
我是 Ryan，方向是 AI 应用开发 / Java 后端 / LLM / RAG / Agent 工作流。
我正在做 简喵：一个 岗位描述（JD）驱动的 AI 求职准备平台。
简喵不是普通 AI 简历模板工具，也不是单纯的 ATS 打分玩具。它围绕一个具体岗位描述，把简历体检、岗位匹配诊断、候选人证据链、证据约束改写、Guard Trail、面试准备包、求职信生成和岗位材料工作台连接成一个闭环。
我更关注的是：如何把 AI 从“能生成文本”推进到“可验证、可追溯、可落地的工程产品”。
详细中文介绍见：README.zh-CN.md