π Dashboard
Total Jobs
To Apply
Applied
Avg Score
Recent Jobs
| Title | Company | Score | Decision | Status | Link |
|---|
πΌ All Jobs
| Title | Company | Score | Decision | Status | Link |
|---|
βοΈ Configuration
Profile Settings
Matching Settings
π CV Generator
Generate CV Based on Job Description
Paste a job description, and the system will select the best CV variant and generate a tailored .docx file for download.
π Activity Logs
π§ Smart CV Generator
[ACTION VERB] + [TECHNOLOGY] + [BUSINESS CONTEXT] + [MEASURABLE IMPACT]
Generate Smart CV
βΉοΈ How the Unified Smart CV Engine Works
1. Role Detection β JD classified into one of 8 roles (4 testing + 4 leadership). Two-pass scorer: title keywords (20 pts) then supplementary signals (3 pts). Tone adjusted per role.
2. 3-Tier JD Alignment β Every bullet scored against the JD: Exact (skill in JD & bullet = 10 pts), Adjacent (same technology domain cluster = 5 pts), Transferable (abstract pattern match = 2 pts).
3. Recency-weighted Selection β Top 4 bullets from the 2 most recent companies, top 3 from older ones. Always relevance-ordered within each company.
4. LLM Bullet Transformation β Rewritten by Ollama with formula:
[ACTION VERB] + [TECHNOLOGY] + [BUSINESS CONTEXT] + [MEASURABLE IMPACT]
LLM is explicitly instructed not to fabricate any metric not in the original.
5. ATS Optimisation β Exact-match JD keywords guaranteed in cover; LLM adds 5 supplementary ATS phrases. Embedded invisibly (white 1pt text) for maximum ATS score.
6. Template β Nilanjan Islam's CV layout preserved in full (fonts, borders, spacing) via lxml in-place modification.