Running Lab
Job Radar
A daily job intelligence system that scans company career pages, ranks fresh roles against candidate profiles, and sends automated email reports.
In a recent run, it checked 112 company sources, evaluated 14,977 roles, and surfaced 6 new recommendations from 530 score-matching jobs.
Problem
Job boards are noisy, stale, and poorly personalized. The most valuable opportunities are often the newest roles that closely match a candidate’s real background.
Approach
- Crawl company career pages daily
- Normalize job data across sources
- Score roles against multiple candidate profiles
- Prioritize fresh P0 / P1 opportunities
- Generate markdown reports
- Send automated email digests
- Prepare the data model for a future dashboard view
System shape
Company career pages
crawler adapters
normalized job records
candidate profile scoring
ranked opportunities
markdown report / email digest / future dashboard
The main design goal is to crawl once, normalize early, and reuse the same scored job records across markdown reports, automated email digests, and future dashboard views.
Sample daily report
A sanitized preview of the markdown report generated from normalized job records and delivered as an automated email digest.
Job Radar Daily Report
Generated May 29, 2026
Pipeline summary
112
Companies checked
14,977
Jobs fetched
14,977
Jobs evaluated
530
Score-matching jobs
260
Filtered out by location
14,708
Skipped
Current backlog
Daily changes
New recommendations 6
Closed matching jobs 5
Failed sources 2
Duplicates hidden 9
Top recommendation
Senior Frontend Engineer, Community Platform
P0Location Remote - United States
Location match US Remote
Recommendation P0 — Apply today
Tags LEVEL_GOOD
Score 222
Reasons
- P0 Apply Today: personal-fit rerank
- Role fit: explicit frontend / web / UI title
- Level fit: senior is strongly aligned
- Strong apply: high score with target title signal
Closed matching jobs
Full Stack Engineer, AI Notebook Product — Example AI Product
Senior Frontend Engineer, AI — Example Fintech Platform
Staff AI Engineer, AI/ML Ops — Example Observability Platform
Failed sources
2 sources failed and were included in the report for crawler maintenance.
Markdown is used as a portable output format, so the same report can be read in email, saved for history, or later rendered inside a dashboard.
Initial profiles
Frontend Systems / Product UI
AI Platform / Infrastructure
AI Agent / Agent SDK
Full-stack / AI Applications
Developer Experience / Internal Tools
Status
Running internally: crawls roles, generates ranked markdown reports, and sends automated email digests.
Next
- Dashboard view for ranked roles
- Profile selector
- Explainable scoring breakdowns
- Agent-assisted profile recommendation
- Multi-recipient email configuration