PM Job Tracker

Paste any careers-site URL → get the Product Manager roles relevant to your resume, with link + latest posting date, newest first. Your resume stays local; nothing is uploaded.

What it does

A personal job-hunt assistant for Product Managers. Give it a careers-site URL — or just run it for a default India-wide sweep — and in one pass it:

1. Fetches the latest openings directly from the source: Workday, Greenhouse, Lever, Ashby and LinkedIn through their own APIs (real posting dates included), and any other careers page by rendering it in a real headless browser.
2. Reads your resume afresh on every run and builds a matching profile from it — your skills (product strategy, roadmaps, agile, analytics, A/B testing) and domains (fintech, telematics, SaaS, AI/LLM). Edit the resume and the matching adjusts on its own; there's no re-import step.
3. Scores and tags each role against that profile — relevance and experience-fit (explained below).
4. Ranks and exports the matches to a formatted Excel — best experience-fit first, then most recent — with clickable links, a fresh file per scan (e.g. Linkedin_17-06.xlsx), and a NEW tag on anything posted since your last run.

Nothing leaves your machine — your resume, settings and results all stay local.

What each column means

Column	What it tells you
Fit%	How well the role matches your résumé (0–100): title relevance (is it really a PM role?) + how many of your résumé's skills/domains the posting mentions. Higher = closer fit.
Exp Fit	How the role's required years compare to your own PM experience (set in config.json), within a ±tolerance. Fits = your level · Stretch (+Ny) = wants N more years · Over-qualified = wants less · Unknown = the posting didn't state it.
Posted	The posting date (exact from APIs, "x days ago" → date from LinkedIn).
Title / Company / Location	The role, employer and city.
Req Exp	The raw experience line pulled from the job description (e.g. 5–8 years).
Source	Which board it came from (linkedin, greenhouse:razorpay, workday:pwc, …).
Link	Clickable link straight to the live posting.

Roles are ordered Fits first → newest date → Fit%, so the most relevant, freshest, right-level roles sit at the top.

How it matches "any" site

Site type	How it's read	Dates?
Workday (*.myworkdayjobs.com — PwC, etc.)	Official JSON API	✅
Greenhouse / Lever / Ashby	Official JSON API	✅
LinkedIn	Public guest jobs API (no login)	✅ (relative)
Any other careers page	Real headless Chrome (Playwright) renders + extracts	when shown

LinkedIn rate-limits aggressively — large/fast pulls may get temporarily blocked. The tool paces requests and falls back to the browser renderer.

Setup (one time)

pip install -r requirements.txt
python -m playwright install chromium
# create your own config from the template, then edit the resume path:
copy config.example.json config.json

Playwright + Chromium are only needed for the generic (non-API) sites. The Workday/Greenhouse/Lever/Ashby/LinkedIn paths work without them.

Usage

# Scan a single URL
python tracker.py scan "https://pwc.wd3.myworkdayjobs.com/en-US/Global_Experienced_Careers"

# Scan a LinkedIn search URL
python tracker.py scan "https://www.linkedin.com/jobs/search?keywords=product%20manager&location=India"

# Only newly-appeared roles since last run
python tracker.py scan "<url>" --new-only

# Loosen / tighten the relevance bar (default 40)
python tracker.py scan "<url>" --min-score 30

# Build a watchlist and scan it all at once
python tracker.py add-source "<url1>"
python tracker.py add-source "<url2>"
python tracker.py scan-all

When you update your resume

Just save the edited file at the same path — the tool re-reads it on every run, so nothing else to do. If the new file lives elsewhere:

python tracker.py set-resume "C:\path\to\your_resume.docx"

Output

• A ranked table in the terminal (Fit%, date, NEW tag, title, location, exp-fit, link).
• A formatted Excel in results\ per scan — named by source + date (Linkedin_17-06.xlsx): bold/centered/coloured headers, colour-coded Exp Fit, clickable links, frozen header and filters on.
• history.json remembers seen jobs so --new-only shows only what's new.

Run with no arguments for the default India-wide sweep:

python tracker.py

Tuning

• Relevance bar: min_score in config.json (or --min-score).
• Keyword weights / penalised titles: profile.py and scorer.py.

Disclaimer

For personal, educational use. It reads public job-board endpoints; respect each site's Terms of Service and rate limits. Your resume, config and results stay local and are git-ignored — they are never committed.