English | 中文

InterviewMaster

An AI interview training platform for individual job seekers. Invited users register with an invitation code, upload a resume, choose a primary technical language and target company, and start a backend interview simulation with dynamic follow-up questions and a complete, scoreable review record.

The M0–M3 and local Beta capabilities are complete, including database migrations, the real HTTP loop, and Whisper audio transcription validation.

The initial MVP focuses on text-based mock interviews. Interview blueprints and candidate questions are internal orchestration resources rather than user-managed assets. Interview intelligence search and ASR are available under /api/v1/beta.

Technical baseline

• Backend: Go, go-zero REST, Asynq Worker, PostgreSQL 18 + pgvector, Redis, and S3-compatible object storage.
• Frontend: Next.js App Router, React, TypeScript, and pnpm.
• Contract: backend/api/interviewmaster.api is the single source of truth for the HTTP API; Go handlers, OpenAPI, and the TypeScript SDK are generated from it.

See the English documentation index. The product line is described in Launch redesign, and the remaining invite-beta work is in Invite beta TODO. The historical technical baseline is in docs/en/archive/.

Local setup

Prepare the environment file:

Copy-Item .env.example .env
docker compose up -d postgres redis minio tika

Run database migrations:

cd backend
$env:IM_DATABASE_URL = "postgres://interviewmaster:interviewmaster_local_only@localhost:5432/interviewmaster?sslmode=disable"
go run github.com/pressly/goose/v3/cmd/goose@v3.27.1 -dir migrations postgres "$env:IM_DATABASE_URL" up

Registration requires an invitation code. See the local setup guide for how to create one after the first migration.

Start the API, Worker, and Web separately:

cd backend; go run -buildvcs=false ./apps/api -f apps/api/etc/interviewmaster.yaml
cd backend; go run -buildvcs=false ./apps/worker -f apps/worker/etc/worker.yaml
pnpm --dir web dev

To view the frontend without starting the backend, use the Mock environment (port 3001, separate from the normal development port 3000):

pnpm mock:web

Any email and password can be used in Mock mode. The upper-right corner displays a Mock badge.

You can also start the complete application stack in containers:

docker compose --profile app up --build

Start the local Whisper Beta ASR service:

docker compose --profile beta up -d asr

Common commands

# Generate code from the API contract and SQL
.\scripts\generate.ps1

# Run backend/frontend checks, tests, and builds
.\scripts\check.ps1

# Install frontend dependencies
pnpm install

After starting the API, visit GET http://localhost:8080/api/v1/health and GET http://localhost:8080/api/v1/ready; the default Web address is http://localhost:3000.

Directory

backend/              Go API, Worker, migrations, and SQL queries
web/                  Next.js frontend
docs/                 Localized documentation under en/ and ch/
scripts/              Windows development scripts
docker-compose.yml    Local dependencies and optional full application stack

---

InterviewMaster（中文）

面向个人求职者的 AI 面试训练平台。受邀用户使用邀请码注册后，上传简历、选择主技术语言与目标公司即可开始后端岗位模拟面试，系统会动态追问并保留可评分、可复盘的完整面试记录。

当前 M0～M3 与 Beta 本地能力均已完成，并通过数据库迁移、真实 HTTP 闭环和 Whisper 音频转写验收。

首发 MVP 聚焦文字模拟面试；题目大纲和候选问题属于后台面试编排，不作为用户管理资源。面经检索与 ASR 位于 /api/v1/beta。

技术基线

• 后端：Go、go-zero REST、Asynq Worker、PostgreSQL 18 + pgvector、Redis、S3 兼容对象存储。
• 前端：Next.js App Router、React、TypeScript、pnpm。
• 契约：backend/api/interviewmaster.api 是 HTTP API 单一事实来源；Go handler、OpenAPI 和 TypeScript SDK 均从它生成。

文档目录见 中文文档索引。产品主线见 上线重构_0814.md，邀请制剩余项见 邀请制todo_0828.md；历史技术基线已移入 docs/ch/archive/。

本地启动

先准备环境文件：

Copy-Item .env.example .env
docker compose up -d postgres redis minio tika

执行数据库迁移：

cd backend
$env:IM_DATABASE_URL = "postgres://interviewmaster:interviewmaster_local_only@localhost:5432/interviewmaster?sslmode=disable"
go run github.com/pressly/goose/v3/cmd/goose@v3.27.1 -dir migrations postgres "$env:IM_DATABASE_URL" up

注册需要邀请码；首次启动后的创建方式见 本地启动指南。

分别启动 API、Worker 和 Web：

cd backend; go run -buildvcs=false ./apps/api -f apps/api/etc/interviewmaster.yaml
cd backend; go run -buildvcs=false ./apps/worker -f apps/worker/etc/worker.yaml
pnpm --dir web dev

只想看前端页面、不启动后端时，用 Mock 环境（端口 3001，与正式开发的 3000 分开）：

pnpm mock:web

任意邮箱和密码都可以登录。右上角会显示 Mock 标识。

也可以启动完整应用容器：

docker compose --profile app up --build

启动本地 Whisper Beta ASR：

docker compose --profile beta up -d asr

常用命令

# 根据 API 契约和 SQL 生成代码
.\scripts\generate.ps1

# 后端和前端静态检查、测试、构建
.\scripts\check.ps1

# 前端依赖安装
pnpm install

运行 API 后可访问 GET http://localhost:8080/api/v1/health 与 GET http://localhost:8080/api/v1/ready；Web 默认地址为 http://localhost:3000。

目录

backend/              Go API、Worker、迁移和 SQL 查询
web/                  Next.js 前端
docs/                 中英文文档分别位于 docs/en/ 与 docs/ch/
scripts/              Windows 本地开发脚本
docker-compose.yml    本地依赖及可选完整应用编排