Problem
Job discovery for wireless roles across multiple platforms was inconsistent and time-heavy. Repeated postings, stale links, and inconsistent metadata made it hard to prioritize high-fit opportunities quickly.
//PROJECT 01
Wireless Job Discovery Pipeline
I built a practical automation pipeline to monitor wireless/network job postings, score fit, and reduce manual searching overhead. The goal was to turn a repetitive daily process into a reliable data workflow with clean outputs.
Constraints
- Source sites with different structures and varying metadata quality
- Need for repeatable daily runs without manual cleanup
- Lightweight local tooling that can run on a personal workstation
Approach
I separated the system into collection, normalization, scoring, and reporting stages. This made it easier to tune ranking logic, maintain source connectors, and isolate failures by stage when sites changed.
Implementation
- Used `requests` and `BeautifulSoup` for source collection and parsing
- Used Playwright where dynamic rendering blocked straightforward scraping
- Normalized records with Pydantic models before persistence in SQLite
- Added stale-link checks and dedup logic to improve result quality
- Generated ranked output for quick daily decision-making
Result
The pipeline transformed job search from a manual browser-heavy routine into a structured workflow. It improved signal quality, reduced duplicate review effort, and created clearer visibility into role fit over time.
What I Would Improve Next
- Add source reliability scoring and retry backoff per connector
- Add better change-tracking to detect meaningful posting edits
- Expose results in a compact dashboard view for trend analysis