hero background

Hi, I'm Xinran Hao

I am shifting from backend development to AI Agent development, with a focus on LLM applications, tool calling, and companion-oriented agents.

Explore My Work
Python
TypeScript
LLM
Agent
Harness Engineering
Tool Calling
MCP
RAG
Web App
Companion Agent

Projects

Selected Work

AI Code Helper For Hot100

A Spring Boot + Vue AI coding assistant built around a self-built ReAct Agent Runtime, with plan-before-act execution, permission-gated tools, hybrid RAG, background tasks, and traceable Hot100 coaching.

Java 21Spring Boot 3.5ReAct AgentHybrid RAGSSELangChain4j
GitHub

Backend Highlights

  • Built a custom ReAct Agent Runtime with a validated plan-before-act stage and up to 8 autonomous model/tool turns.
  • Implemented permission-gated tools across READ, WRITE, EXTERNAL, and SENSITIVE levels with structured tool_result recovery instead of hard failures.
  • Added Redis-backed tool idempotency using SHA-256 hashes to detect repeated tool calls and replay cached results during retries.
  • Designed three-tier context compaction, tiered tool recovery, hook events, and step-level SSE streaming for observable agent execution.
  • Modeled AgentTask, RuntimeSlot, and AgentStep so retries, runtime traces, background tasks, and executor watchdogs can be inspected clearly.
  • Built hybrid explainable RAG for Hot100 resources by combining Redis vector similarity with deterministic keyword scoring through HybridRanker.
  • Added long-term memory for user preferences, weaknesses, wrong answers, next actions, and notes.
  • Used Spring Boot 3.5, Java 21, Spring Security JWT, JPA, Flyway, Redis, LangChain4j, DashScope/Qwen, and Docker Compose.

Full Tech Stack

Java 21Spring Boot 3.5Spring Security JWTVue 3LangChain4jDashScope/QwenMCPSSEHybrid RAGRedis Vector StoreMySQLRedisFlywayDocker Compose

CityLife Review

A Spring Boot 3.4 backend for a local life-review platform, combining Redis business patterns, Elasticsearch search, RabbitMQ flash-sale ordering, and a Spring AI RAG recommendation agent.

Java 17Spring Boot 3.4Spring AIRedis StackElasticsearchRabbitMQ
GitHub

Backend Highlights

  • Implemented phone-code login, Redis token storage, shop caching, GEO nearby search, ZSet likes/feed, Set follows, and Bitmap sign-in.
  • Built high-concurrency voucher ordering with Redis Lua stock checks, one-user-one-order constraints, pre-deducted stock, and purchase qualification records.
  • Designed RabbitMQ async order creation with persisted PROCESSING state, publisher confirm, return callback, manual ACK, DLQ, Redis rollback, and stale-order compensation.
  • Used Redisson distributed locks, RRateLimiter-based declarative rate limiting, and MySQL unique indexes for concurrency control.
  • Added Elasticsearch full-text shop search with geo-distance sorting and structured shop document indexing.
  • Built a Spring AI recommendation agent with ChatClient, tool calling, chat memory, and preference extraction stored in Redis.
  • Implemented semantic review search through a RAG pipeline with multi-query expansion, HyDE, vector retrieval, LLM reranking, and sentiment tagging.
  • Documented APIs with SpringDoc OpenAPI and added CI/CD support through GitHub Actions.

Full Tech Stack

Java 17Spring Boot 3.4.5Spring AIMyBatis-PlusMySQL 8Redis StackRedissonRabbitMQElasticsearchRAGLuaSpringDoc OpenAPIGitHub Actions

Personal Portfolio

A personal website built to showcase projects, technical skills, learning journey, and contact information.

Next.jsTailwind CSSshadcn/uiVercel
GitHub

Key Features

  • Responsive personal portfolio homepage
  • Project showcase section
  • Skills and learning journey sections
  • Light / dark theme support
  • Smooth scroll and subtle animations

What I Built

  • Built a modern portfolio website with Next.js and Tailwind CSS.
  • Designed sections for projects, skills, about, journey, and contact.
  • Added theme switching, scroll animations, and back-to-top interaction.

What I Learned

  • Learned how to structure a personal portfolio website.
  • Practiced responsive layout and component-based frontend development.
  • Improved UI polish through spacing, typography, and animation tuning.

Learning Resources

Learning Resources

A curated learning map for my Java backend foundation and AI Agent development path.

Backend Fundamentals

JavaGuide

A structured Java backend guide for interviews and system design.

JavaInterviewSystem DesignDatabase

小林 coding

Visual notes for CS fundamentals, MySQL, Redis, networks, OS, and backend interviews.

NetworkOSMySQLRedis

AI Agent Learning

Hello Agents

A Chinese AI Agents tutorial focused on core principles, agent architecture, classic paradigms, and building AI Native Agents.

AI AgentArchitecture

Hugging Face Agents Course

A hands-on agents course covering agent fundamentals, tools, smolagents, LangGraph, LlamaIndex, Agentic RAG, observability, and final projects.

AI AgentToolsAgentic RAG

Skills

Technical Focus

LLM Application × Agent Development

A compact map of the technical directions I keep learning and practicing across LLM applications, AI agents, tool calling, and web product delivery.

01

Agent Core

I focus on agent task planning, tool calling, memory mechanisms, and multi-turn interaction capabilities.

AI AgentTool CallingFunction CallingMCPRAGMemoryPlanning
02

LLM Application Engineering

I practice model API integration, prompt design, context management, and application integration in projects.

LLM APIPrompt EngineeringContext EngineeringLangChainDashScopeClaude APIOpenAI APIEmbedding
03

Web Productization

I want to turn AI capabilities into usable web applications, especially companion agents and human-AI interaction experiences.

PythonTypeScriptVue 3Web AppFastAPIDockerGitCompanion Agent

About

About Me

I am an undergraduate student majoring in Artificial Intelligence, shifting from traditional backend development toward LLM applications and AI Agent development. Rather than only implementing business APIs, I care more about combining large language models, tool calling, and web products into intelligent applications that are genuinely useful.

I am currently learning and practicing RAG, Tool Calling, MCP, Prompt Engineering, and Agent workflows, and I am especially interested in companion agents, human-AI interaction, and emotionally supportive AI products.

Learning Journey

From Engineering Foundations to AI Agents

2024

Engineering Foundations

Studied Java Web, databases, and basic backend development to build an engineering understanding of web applications.

2025

Project Practice

Practiced API design, database design, deployment, and collaborative development through projects such as a campus food ordering system.

2026

AI Agent Exploration

Shifted toward LLM applications and Agent development, exploring RAG, tool calling, MCP, and companion-oriented agent products.

Contact

Let's Connect

I am looking for internship opportunities related to AI Agents, LLM application development, or AI product engineering, and I am also open to project collaboration and technical discussions.

Email MeGitHubDownload Resume