📡 Nokrify — Pakistan Data Jobs Radar

A modern, persistent web application and radar built specifically for fresh graduates in BSc Computer Systems Engineering to automatically monitor any data-related role across Pakistan, listed within the last 48 hours.

Aggregates directly from LinkedIn Pakistan, Rozee.pk, and Indeed, checks for cross-board duplicates, calculates fit scores against your specific engineering skill set, and provides a sleek dashboard with 100% real, verified, working application URLs.

---

🎯 Candidate Profile & Matching Model

Calibrated for your exact fresh graduate profile:

• Degree: BSc Computer Systems Engineering (Fresh Graduate, Pakistan)
• Core Skills: SQL (Complex queries, CTEs, Window functions), Power BI (DAX, Power Query), Excel (Pivot tables, Advanced formulas), MySQL, PostgreSQL, Microsoft Azure (Azure Data Factory, ADLS Gen2, Azure SQL), AWS (S3, AWS Glue), Python (Pandas, PySpark), Git/GitHub.
• Certifications: Google Data Analytics Certificate, IBM Data Analytics Professional Certificate.
• Target Seed Roles: Data Analyst, Data Engineer, Data Scientist, BI Analyst/Developer, ETL Developer, Database Administrator (DBA), Reporting Analyst, MIS Officer/Executive, Data Operations, Business Intelligence, Analytics Associate, Graduate Trainee.
• Scope: Open to all industries across Pakistan (Karachi, Lahore, Islamabad/Rawalpindi, Remote, etc.).

Weighted Scoring Algorithm (0–100%):

1. Core Skills Match (50 pts max): Scans for exact technical stack matches with boundary-safe phrase analysis.
2. Target Role & Title Match (30 pts max): Checks job title against seed role keywords.
3. Fresh Graduate / Seniority Fit (15 pts max): Rewards entry-level, fresh graduate, intern, trainee, and 0–2 year postings (+15 pts bonus); penalizes 5+ year senior roles.
4. Degree & Engineering Fit (5 pts max): Validates alignment with Computer Systems Engineering, Computer Science, and IT degrees.

---

⚡ Key Upgrades & Capabilities

1. Direct LinkedIn Pakistan Integration (app/scrapers/linkedin.py)

• Leverages LinkedIn's public guest job search API with standard rotating headers.
• Enforces strict 48-hour freshness filtering (f_TPR=r172800) to guarantee you only see active postings from today or yesterday.
• Extracts real, verified https://pk.linkedin.com/jobs/view/... URLs that open directly on LinkedIn.

2. Strict 48-Hour Freshness Controls

• Default 48h Filter: Only listings posted in the last 48 hours are shown by default.
• Quick Age Horizon Toggle: Switch with one click between Last 24h, Last 48h, Past Week, and All Time.
• Every posting displays an animated green freshness badge (e.g. ⚡ 10h ago, ⚡ Yesterday).

3. Beautiful High-End Web Interface

• Modern Obsidian & Emerald Aesthetic: Dark mode with radial glows, glassmorphism cards, and crisp typography.
• Dynamic Company Avatars: Distinct visual brand initials with gradient color palettes.
• Match Score Gauges: Glowing status pills displaying fit percentage and complete scoring breakdown on click.
• Bento Stats Grid: Live counters for 48h Fresh Roles, Top Fit (≥70%), LinkedIn Roles, Applied, Interviewing, and Active Sources.
• One-Click Actions: Quick status dropdown (New, Viewed, Applied, Interviewing, Rejected), "Copy Link" button with toast notification, and direct "Apply on Official Board" button.

4. Fuzzy Cross-Board Deduplication (app/services/deduplication.py)

• Merges duplicates when a company posts the same job across LinkedIn, Rozee, and Indeed.
• Displays a Boards Merged badge with all alternative links preserved in the details drawer.

5. Resume & Skill Profile Tuning Drawer (app/services/resume_parser.py)

• Upload an updated PDF or DOCX resume to extract new skills.
• Manually add or remove skills with instant live re-scoring of all 48h jobs in the database.

---

🚀 How to Run

Windows (One-Click)

Double-click run.bat in the project root:

run.bat

Command Line

python run.py

Then navigate to:

http://localhost:8000

---

🧪 Running Automated Tests

Run the full pytest suite:

python -m pytest -v

Verifies API endpoints, LinkedIn scraper, fuzzy deduplication, and matching engine scoring.