How chenxin1209/fusioncareer-algorithm works
FusionCareer-Algorithm combines job-post structuring and deduplication, resume parsing, and conversational job recommendation. The main job workflow turns Markdown articles into structured, deduplicated CSV/JSON; resume inputs are extracted to text and parsed into structured JSON; recommendation gathers user preferences and returns ranked job results. The README also documents an admin parsing service and optional backend upload. Sampled source confirms only parts of these paths, so uncertain service wiring is omitted.
Main entry points: parser.py Resume parser · run_pipeline.py Batch pipeline · engine.py Extract and deduplicate · serve.py Recommendation service
Components 26
Job structuring
- Batch pipeline
pipeline/run_pipeline.py - Extract and deduplicate
job_structuring/engine.py - Keyword prefilter
job_structuring/prefilter.py - Positions CSV
- JSON export
job_structuring/export.py - Positions JSON
Resume parsing
- Resume parser
resume_parser/parser.py - File extractors
resume_parser/extractors - Image OCR
resume_parser/ocr - DeepSeek client
resume_parser/llm/deepseek_client.py - Structured resume
Job recommendation
- Recommendation service
job_recommend/serve.py - Preference dialogue
job_recommend/dialogue.py - Filter query
job_recommend/filters.py - Job ranking
job_recommend/rank.py
Service and export
- Admin parsing
job_structuring/admin_parse.py - Job service
job_structuring/serve.py - Backend upload
pipeline/upload_to_backend.py - Resume API
resume_parser/main.py - Resume routes
resume_parser/routers
Other components
- Markdown articles
- LLM service
- Java backend
- Resume submitter
- DeepSeek API
- Job seeker
Connections 22
- Markdown articles → Batch pipeline: provides articles
- Batch pipeline → Extract and deduplicate: runs batch
- Extract and deduplicate → Keyword prefilter: checks keywords
- Extract and deduplicate → LLM service: extracts fields
- Extract and deduplicate → Positions CSV: writes rows
- Batch pipeline → Positions CSV: deduplicates
- Batch pipeline → JSON export: exports CSV
- JSON export → Positions JSON: writes JSON
- Admin parsing → LLM service: parses fields
- Backend upload → Java backend: uploads positions
- Resume submitter → Resume API: submits resume
- Resume API → Resume routes: dispatches requests
- Resume routes → Resume parser: parses resume
- Resume parser → File extractors: extracts text
- Resume parser → Image OCR: reads images
- Resume parser → DeepSeek client: sends text
- DeepSeek client → DeepSeek API: requests JSON
- Resume parser → Structured resume: normalizes output
- Job seeker → Recommendation service: requests matches
- Recommendation service → Preference dialogue: collects preferences
- Preference dialogue → Filter query: builds query
- Recommendation service → Job ranking: ranks jobs