Recruitimation

AI-powered candidate screening for light industrial staffing.

Ingests applicants from Indeed / LinkedIn via a Chrome extension, scores them with embeddings + Claude, and presents them in a swipe-style review dashboard.

---

Architecture

Chrome Extension  →  POST /api/ingest  →  Backend (Fastify + Prisma)
                                                │
                                         BullMQ queue
                                                │
                                      AI Scoring Worker
                                       ├── OpenAI embeddings (match score)
                                       └── Claude (willingness score + summary)
                                                │
                                         PostgreSQL + R2
                                                │
                                      React frontend (swipe deck)

---

Quick Start

Prerequisites

• Node 20+
• Docker (for Postgres + Redis)

1. Start infrastructure

docker-compose up -d

2. Backend

cd backend
cp .env.example .env        # fill in API keys
npm install
npm run db:push             # apply schema to Postgres
npm run dev                 # API on :3000

In a second terminal:

npm run worker              # BullMQ scoring worker

3. Frontend

cd frontend
npm install
npm run dev                 # Vite on :5173

4. Chrome Extension

1. Open chrome://extensions
2. Enable Developer mode (top right)
3. Click Load unpacked → select the extension/ folder
4. Pin the extension and set the API URL to http://localhost:3000

---

Phase 1 — Paste & Parse

Visit http://localhost:5173/parse to paste a candidate profile and validate AI scoring without the extension.

---

Environment Variables

Key	Description
DATABASE_URL	Postgres connection string
REDIS_URL	Redis connection string
OPENAI_API_KEY	For text-embedding-3-small
ANTHROPIC_API_KEY	For Claude (scoring + outreach)
R2_ACCOUNT_ID	Cloudflare R2 account ID
R2_ACCESS_KEY_ID	R2 access key
R2_SECRET_ACCESS_KEY	R2 secret key
R2_BUCKET_NAME	R2 bucket name (default: recruiting-pdfs)

---

Project Structure

recruitimation/
├── backend/            Node.js + Fastify API + BullMQ worker
│   ├── prisma/         Database schema
│   └── src/
│       ├── routes/     API endpoints
│       ├── workers/    BullMQ scoring worker
│       ├── ai/         Embeddings, scoring, draft generation
│       └── storage/    Cloudflare R2
├── frontend/           React + Vite + Tailwind
│   └── src/
│       ├── views/      SwipeDeck, ResultsGrid, OutreachFlow, JobManager, PasteAndParse
│       ├── components/ CandidateCard, ScoreChip, PdfViewer
│       └── api/        API client
└── extension/          Chrome Manifest V3
    ├── content/        indeed.js, linkedin.js
    ├── popup/          Extension popup UI
    ├── interceptor.js  LinkedIn XHR/fetch proxy (page context)
    └── background.js   Service worker

---

Build Phases

Phase	Status	Description
1	✅ Done	Paste & Parse MVP — validate AI scoring
2	✅ Done	Core API, DB schema, swipe deck
3	✅ Done	Chrome extension (Indeed + LinkedIn)
4	✅ Done	Outreach draft + confirm flow
5	✅ Done	ATS logging + CSV export
6	🔲 Next	LinkedIn DOM hardening, pin remind cron
7	🔲 Next	PCF PDF generation
8	🔲 Next	Auth (Clerk / Auth.js)