{"id":"c979202d-157d-4ae9-9d6b-5187ff738884","entityType":"agent","slug":"crewai-vamsiraju6363-job-application-crew","name":"Job_application_crew","canonicalUrl":"https://www.xpersona.co/agent/crewai-vamsiraju6363-job-application-crew","canonicalPath":"/agent/crewai-vamsiraju6363-job-application-crew","generatedAt":"2026-10-09T01:05:28.720Z","source":"GITHUB_OPENCLEW","claimStatus":"UNCLAIMED","verificationTier":"NONE","summary":{"evidence":{"source":"editorial-content","verified":true,"confidence":"high","updatedAt":"2026-05-18T06:45:12.107Z","emptyReason":null},"description":"Multi-agent job-discovery tool built with CrewAI, Serper, and OpenAI. Ranks job listings against a resume via semantic matching. Job Application Crew A multi-agent system that does the boring parts of a job search for you. It searches, scrapes, reads job descriptions, and ranks them against your resume using semantic matching — so you can focus on the roles that actually fit. Built with $1, Serper for search, and OpenAI for the heavy lifting. Notebook-first so you can see every step and tune it for your own pipeline.

\"Python\"\n \"Python\"\n \"CrewAI\"\n \"OpenAI\"\n \"Jupyter\"\n

\n\n---\n\n## Why I built this\n\nApplying to jobs is mostly a parsing problem. Every posting looks the same but describes different work. I wanted a small agent system that could:\n\n1. **Search** job boards for relevant openings,\n2. **Read** the actual job descriptions (not just titles),\n3. **Compare** them to a resume semantically, not with keyword matching, and\n4. **Rank** them so I start with the best-fit roles.\n\nThis is that system.\n\n## How the agents are wired up\n\n```\n ┌───────────────────────────────────────────────┐\n │ CrewAI Crew │\n │ │\n query → │ [ Searcher ] → [ Reader ] → [ Matcher ] │ → ranked list\n │ ↓ │\n │ Serper API │\n │ │\n └───────────────────────────────────────────────┘\n```\n\n| Agent | Role |\n|---|---|\n| **Searcher** | Runs targeted queries through Serper and collects candidate postings. |\n| **Reader** | Scrapes the posting pages, extracts structured JD content, and strips navigation noise. |\n| **Matcher** | Embeds both the JD and the resume, scores similarity, and explains why the role fits. |\n\n## Running it\n\n**Requirements**\n- Python 3.11+\n- Serper API key — [serper.dev](https://serper.dev)\n- OpenAI API key — [platform.openai.com](https://platform.openai.com)\n\n**Setup**\n\n```bash\ngit clone https://github.com/vamsiraju6363/Job_application_crew.git\ncd Job_application_crew\n\npip install -r requirements.txt # notebook includes a pip install cell if you prefer\n\nexport SERPER_API_KEY=\"sk-...\"\nexport OPENAI_API_KEY=\"sk-...\"\n\njupyter notebook Job_application_crew.ipynb\n```\n\nDrop your resume as markdown in the notebook cell that asks for it. Tune the search query. Run all cells. The crew returns a ranked list with a short rationale per role.\n\n## Output shape\n\n```json\n[\n {\n \"title\": \"Senior ML Engineer\",\n \"company\": \"Acme AI\",\n \"url\": \"https://jobs.acme.ai/...\",\n \"score\": 0.87,\n \"why\": \"Strong overlap on RAG, LangChain, and production deployment on AWS.\"\n }\n]\n```\n\n## Tuning knobs\n\n- **Search depth** — how many SERP results to pull per query.\n- **Reader timeout** — gives up on slow pages so the crew doesn't stall.\n- **Matcher threshold** — drop anything scored below it so you don't drown in noise.\n- **Model choice** — swap the OpenAI model per agent. A cheaper model for Reader, a stronger one for Matcher.\n\n## Roadmap\n\n- [ ] Swap Serper for a lightweight board-specific scraper (LinkedIn, Wellfound, Greenhouse).\n- [ ] Add an Applier agent that drafts cover-letter seeds per top match.\n- [ ] Persist results so I can diff against yesterday's run.\n- [ ] Small CLI wrapper so it's not notebook-only.\n\n---\n\nBuilt by [Vamshi](https://www.linkedin.com/in/vamsi-raju/) · ideas, issues, and PRs welcome.\n","readmeExcerpt":"Job Application Crew A multi-agent system that does the boring parts of a job search for you. It searches, scrapes, reads job descriptions, and ranks them against your resume using semantic matching — so you can focus on the roles that actually fit. Built with $1, Serper for search, and OpenAI for the heavy lifting. Notebook-first so you can see every step and tune it for your own pipeline.

\"Python\"