| title | TalentLens AI |
|---|---|
| emoji | 🧠 |
| colorFrom | blue |
| colorTo | purple |
| sdk | docker |
| app_file | app.py |
| app_port | 8501 |
| pinned | false |
Try TalentLens AI here: talentlensai on Hugging Face
Note: For the best experience in the live demo, please upload a smaller candidate dataset (up to a few thousand records). The application is designed to scale to much larger datasets (up to 100,000 candidates), but large files may take longer to process on the free Hugging Face hosting environment.
Explainable Multi-Signal Candidate Intelligence Engine
TalentLens AI ranks candidates the way an experienced recruiter would — combining semantic relevance, skill coverage, career alignment, availability signals, and platform trust indicators instead of relying solely on keyword matching.
TalentLens AI helps recruiters quickly identify the strongest candidates from large talent pools.
The system:
- Reads a Job Description (JD)
- Extracts required skills and role requirements
- Analyzes thousands of candidate profiles
- Computes multi-signal fit scores
- Explains strengths and gaps
- Generates recruiter-ready ranked shortlists
Designed for large-scale hiring workflows, TalentLens AI can process and rank 10,000+ candidates within seconds.
Uses TF-IDF semantic similarity to measure how closely a candidate profile aligns with the job description.
Measures coverage of required JD skills and identifies missing competencies.
Evaluates:
- Years of experience
- Title alignment
- Career progression
- Seniority suitability
Considers:
- Open-to-work status
- Notice period
- Recruiter responsiveness
- Recent activity
Analyzes platform-level indicators such as:
- Response rate
- Profile completeness
- Engagement signals
Each candidate receives:
- Final score
- Hiring recommendation
- Skill match percentage
- Risk flags
- Top strengths
- Missing skills
TalentLens AI combines five independent signals.
| Signal | Purpose |
|---|---|
| Semantic Match | Similarity between JD and candidate profile |
| Skill Match | Coverage of required skills |
| Career Fit | Experience and title alignment |
| Availability | Hiring readiness and responsiveness |
| Platform Signals | Trust and engagement indicators |
The final ranking score is generated using a weighted hybrid scoring model.
Candidates are automatically classified into recruiter-friendly categories:
| Recommendation | Meaning |
|---|---|
| Interview Immediately | Strong overall fit |
| Keep in Pipeline | Good fit for future rounds |
| Review Manually | Requires recruiter review |
| Reject | Low relevance to role |
- Python
- NumPy
- Pandas
- Scikit-learn
- TF-IDF Vectorization
- Cosine Similarity
- Streamlit
- Plotly
- JSON
- JSONL
- CSV
- XLSX
- DOCX
- TXT
- JSON
- Direct text input
- JSONL
- JSON
- CSV
- XLSX
Large candidate datasets are not included in this repository. Users can upload their own datasets directly through the Streamlit interface.
Includes:
- Rank
- Candidate ID
- Name
- Current Title
- Experience
- Final Score
- Semantic Score
- Skill Score
- Career Score
- Availability Score
- Platform Score
- Skill Match Percentage
- Hiring Recommendation
- Risk Flags
- Tier Classification
Top 100 candidates with:
- Candidate ID
- Rank
- Score
- Reasoning
Suitable for recruiter review and hackathon evaluation.
TalentLens-AI/
│
├── app.py
├── scorer.py
├── rank.py
├── jd_parser.py
├── candidate_parser.py
├── embeddings.py
│
├── screenshots/
│ ├── 1dashboard-upload.png.png
│ ├── 2results-overview.png.png
│ ├── 3candidate-details.png
│ ├── 4ranked-table.png.png
│ ├── 5score-analysis.png.png
│ └── 6download-results.png.png
│
├── outputs/
├── requirements.txt
└── README.md
| Candidate Count | Runtime |
|---|---|
| 1,000 | ~1–2 sec |
| 10,000 | ~5–10 sec |
| 50,000 | ~20–30 sec |
| 100,000 | ~40–60 sec |
Performance depends on hardware and dataset size.
- Resume Screening
- Candidate Ranking
- Talent Discovery
- Recruiter Intelligence
- Hiring Analytics
- Talent Pipeline Prioritization
TalentLens AI reduces recruiter screening effort from thousands of profiles to an explainable ranked shortlist within seconds.
Instead of manually reviewing every profile, recruiters can focus on the highest-potential candidates first.
git clone https://github.com/manshi-n/TalentLens-AI.git
cd TalentLens-AI
pip install -r requirements.txt
streamlit run app.py- Resume Parsing
- Candidate Skill Gap Learning Paths
- Recruiter Copilot
- Diversity Analytics
- Interview Question Recommendations
- Advanced Semantic Embeddings
- Real-Time Talent Search
Manshi Negi B.Tech CSE (AI & Data Science) Graphic Era Hill University
Built for Data & AI Hackathon 2026.