AI Career Agent Platform

Language: English | 简体中文

AI Career Agent Platform is an AI-assisted workspace for job search management, job analysis, resume maintenance, and interview preparation. The project is organized as a monorepo and currently focuses on building the core loop from job management to asynchronous background processing. Over time, it is intended to evolve into a deployable personal or team-oriented career workspace.

Features

• Job management: create jobs manually, import jobs from URLs, and display structured job descriptions.
• Application lifecycle: track application status, status history, and job-search funnel metrics.
• Resume management: upload resume files and extract text through asynchronous background tasks.
• Knowledge base foundation: model and index local Markdown or Obsidian vault sources.
• Task system: process background jobs with BullMQ and track task status.
• Interview preparation models: reserve domain models for interview sessions, turns, and evaluation reports.
• Development authentication: uses a development user for now; real authentication is planned for a later phase.

Tech Stack

Layer	Technologies
Web	Next.js 16, React 19, Server Components, Server Actions
API	tRPC, Zod
Database	Prisma, SQLite
Queue	BullMQ, Redis
Testing	Vitest
Tooling	pnpm workspace, TypeScript, ESLint, Prettier

Architecture

The project follows a layered architecture. Pages do not access the database directly:

page / component
  -> tRPC / Server Action
  -> service
  -> repository
  -> Prisma
  -> SQLite

Asynchronous task flow:

page / API
  -> BullMQ queue
  -> task_runs
  -> worker
  -> repository
  -> Prisma
  -> SQLite

Workspace Layout

apps/
  web/                  Next.js web application

services/
  crawler/              Job URL crawling and parsing service
  worker/               BullMQ background worker

packages/
  api/                  tRPC routers, schemas, and services
  auth/                 Development user and authentication context
  config/               Environment variable validation
  core/                 Domain types, status enums, and DTOs
  db/                   Prisma schema, migrations, seed data, repositories
  knowledge/            Knowledge base parsing and chunking
  queue/                Queue names, payload schemas, enqueue helpers
  ai/                   LLM provider and AI capability boundary

Running and Deployment

For installation, initialization, and source-based deployment instructions, see:

RUNNING_GUIDE.md

Development Status

The project is still in an early development stage and currently targets local development and source-based deployment validation. The following foundations are in place:

• monorepo project structure;
• Prisma + SQLite persistence;
• tRPC API boundary;
• core page skeletons for Jobs, Resumes, Knowledge, and Tasks;
• Crawler and Worker service boundaries;
• baseline tests and CI quality checks.

Roadmap

• Improve job-match analysis and resume optimization suggestions.
• Add application kanban views and follow-up reminders.
• Introduce real authentication and multi-user permission boundaries.
• Expand interview preparation, knowledge retrieval, and AI review capabilities.
• Support more complete production deployment options.

Contributing

Before submitting changes, run:

pnpm check

Recommended collaboration boundaries:

• Pages should not access the database directly.
• Business rules belong in the service layer.
• Database access belongs in the repository layer.
• API inputs should be validated with Zod.
• User data must be isolated by userId.
• Long-running tasks should go through BullMQ + worker.

License

This repository does not declare an open-source license yet. Add a LICENSE file before public release.