TA_AI_SaaS — AI-assisted talent acquisition

This repository contains X Crewter / TA_AI_SaaS, a Django-based Talent Acquisition platform with server-rendered UI (DTL), REST APIs, real-time updates over WebSockets, background work via Celery, and an optional standalone AI analysis service that runs LangGraph workloads (often on a separate GPU host) and talks to the web app over HTTP.

Repository layout

Path	Purpose
TI_AI_SaaS_Project/	Main Django project (manage.py, x_crewter/, apps/). This is where you install dependencies and run the app.
TI_AI_SaaS_Project/services/	Standalone AI service (Django + DRF, Redis, LangGraph/Ollama). See TI_AI_SaaS_Project/services/README.md.
docs/	Additional API and user documentation (bulk upload, analysis API, guides).
specs/, .specify/	Specification and planning assets.

The Git root folder name is TA_AI_SaaS; the runnable application code lives under TI_AI_SaaS_Project/.

What the main application does

• Accounts: Registration, activation, login, password reset, JWT in HTTP-only cookies (with optional header-based JWT for migration), session timeout, RBAC middleware, Google / LinkedIn / Microsoft social auth (env-driven).
• Jobs: Job listings, dashboard for subscribers, screening questions, application links.
• Applications: Public apply flow, resume upload, bulk upload, duplication detection, throttles.
• Analysis: Initiate/cancel/rerun bulk AI analysis, results and progress; integrates with the standalone service and Channels for live updates.
• Subscription: Landing and subscription-oriented routes for non-subscribed users.

Architecture (high level)

flowchart LR
  subgraph web["Django web app :8000"]
    UI[DTL + shadcn_django UI]
    API[REST APIs]
    WS[Channels WebSockets]
  end
  Redis[(Redis)]
  subgraph ai["AI service :9000 optional"]
    LG[LangGraph / Ollama]
  end
  UI --> API
  API --> Redis
  WS --> Redis
  API -->|"REST + API key"| ai
  ai -->|"HMAC webhooks"| API
  ai --> Redis

Loading

For local development, run the web app and (if you need real analysis runs) the AI service and Ollama on the same machine; Redis URLs must stay aligned between x_crewter/settings.py and services/config/settings.py.

Technology stack (main project)

Versions are pinned in TI_AI_SaaS_Project/requirements.txt.

• Runtime: Python 3.11+ (see project constraints in code; LangChain notes exist for newer Python).
• Web: Django 5.2.x, Django REST Framework, Django Channels (ASGI WebSockets), django-cors-headers, django-csp.
• Auth: djangorestframework-simplejwt, Djoser, social-auth-app-django, Argon2 password hashing.
• Async / tasks: Celery 5.4, Redis 7.x (broker, Channels layer, shared analysis progress helpers).
• AI (in-app client + graphs): LangChain 1.x, LangGraph 1.x; document parsing uses pypdf, python-docx.
• UI: Django templates, Tailwind CSS (Play CDN in templates), shadcn-django; optional local Tailwind tooling via package.json / tailwind.config.mjs.
• Storage: Local filesystem by default; Amazon S3 or Google Cloud Storage via django-storages when configured.
• Testing: Python unittest via Django’s test runner; Selenium + Chrome for E2E flows; webdriver-manager is listed for driver management.

Prerequisites

• Python 3.11+
• Redis (same URL for Celery, Channels, and analysis-related Redis usage)
• Mail: Default settings in x_crewter/settings.py point SMTP to 127.0.0.1:1025 (typical for MailHog or similar); adjust for your environment.
• AI service path: Ollama (or compatible) for the standalone service — defaults in settings use http://localhost:11434 and a phi4-mini-style model name.
• E2E tests: Google Chrome (or Chromium) available on the machine running Selenium tests.

Quick start (main Django app)

From the repository root:

cd TI_AI_SaaS_Project
python -m venv .venv
# Windows PowerShell:
.\.venv\Scripts\Activate.ps1
# Linux/macOS:
# source .venv/bin/activate

pip install -r requirements.txt
cp .env.example .env
# Windows (PowerShell): Copy-Item .env.example .env
# Edit .env: SECRET_KEY, ALLOWED_HOSTS, REDIS_URL, STORAGE_BACKEND, OAuth secrets if used, etc.

python manage.py migrate
python manage.py runserver 0.0.0.0:8000

runserver is extended by Channels for ASGI, so WebSockets work in development without a separate process.

Celery (background tasks)

Celery uses the same REDIS_URL as the rest of the app. On Windows the project sets CELERY_WORKER_POOL = 'solo'. Example:

cd TI_AI_SaaS_Project
celery -A x_crewter worker -l info
celery -A x_crewter beat -l info

Standalone AI service (optional)

Use this when exercising bulk analysis end-to-end or developing the GPU-side worker.

cd TI_AI_SaaS_Project
pip install -r services/requirements.txt
cp services/.env.example services/.env
# Windows: Copy-Item services\.env.example services\.env
# Align API_KEYS / WEBHOOK_SECRET with Django's AI_SERVICE_* settings; see services/README.md

python services/manage.py runserver 0.0.0.0:9000

Full detail: TI_AI_SaaS_Project/services/README.md.

Environment variables (main app)

Copy TI_AI_SaaS_Project/.env.example to .env. Commonly adjusted values:

• SECRET_KEY, DEBUG, ALLOWED_HOSTS
• REDIS_URL — shared with Celery, Channels, and AI progress
• CORS_ALLOWED_ORIGINS, CSRF_TRUSTED_ORIGINS — especially if you use a LAN IP or a separate origin
• STORAGE_BACKEND — local (default), s3, or gcs, plus cloud credentials when not local
• Social auth: GOOGLE_OAUTH2_*, LINKEDIN_OAUTH2_*, MICROSOFT_GRAPH_*
• AI integration: AI_SERVICE_BASE_URL, AI_SERVICE_API_KEY, AI_SERVICE_WEBHOOK_SECRET, AI_SERVICE_WEBHOOK_TOLERANCE_SECONDS, timeouts and circuit breaker thresholds (see x_crewter/settings.py)
• LLM defaults: OLLAMA_BASE_URL, OLLAMA_MODEL

Optional attributes such as FRONTEND_URL / BACKEND_URL are referenced in account flows for absolute links; add them in settings if you need non-default URLs in emails or redirects.

API and UI entry points

• Health: GET /api/health/
• Accounts API: under /api/accounts/
• Applications API: under /api/applications/
• Analysis API: under /api/analysis/; analysis UI under /analysis/
• Dashboard: /dashboard/ (jobs)
• Public apply routes: /apply/, /application/ (see URL conf in x_crewter/urls.py)

Testing

Run the full suite from TI_AI_SaaS_Project:

python manage.py test

• Selenium E2E tests under apps/*/tests/e2e/ require Chrome and may be slower; run a subset when iterating, e.g. python manage.py test apps.accounts.tests.e2e.
• Coverage: A helper script exists at apps/accounts/tests/test_coverage.py; it expects the coverage package (pip install coverage). Project rules target high line coverage for new work.

Service-only tests:

python services/manage.py test services.tests

Documentation

• docs/api/analysis_api.md
• docs/api/bulk_upload.md
• docs/applications-api.md
• docs/user/ai_analysis_guide.md
• AI service: TI_AI_SaaS_Project/services/README.md

License

MIT License — Copyright (c) 2025 MadsGit-98

Contributing / clone

git clone https://github.com/MadsGit-98/TA_AI_SaaS.git
cd TA_AI_SaaS

All application commands assume the working directory is TA_AI_SaaS/TI_AI_SaaS_Project unless noted.