TalentForge is an AI-powered candidate ranking system developed for the India Runs Data & AI Challenge.
Traditional ATS platforms rely heavily on keyword matching, which often overlooks qualified candidates and promotes profiles optimized for keywords rather than actual suitability.
TalentForge evaluates candidates using a combination of:
- Semantic relevance
- Professional experience
- Behavioral hiring signals
- Profile consistency checks
The goal is to generate a recruiter-quality shortlist rather than a keyword-based ranking.
Recruiters review hundreds of profiles for a single role.
Most automated systems focus on keyword overlap and fail to understand:
- Candidate quality
- Hiring likelihood
- Career progression
- Profile authenticity
This challenge requires building an intelligent ranking system that identifies candidates who genuinely fit the job requirements.
The ranking system evaluates candidates using four major dimensions:
Measures alignment between candidate profiles and job requirements.
Examples:
- Retrieval systems
- Search infrastructure
- Ranking systems
- Embedding technologies
- Recommendation systems
Evaluates:
- Years of experience
- Role seniority
- Career progression
Utilizes platform activity signals such as:
- Recruiter response rate
- Interview completion rate
- Profile completeness
- GitHub activity
- Search appearance
Detects suspicious or inconsistent profiles by comparing:
- Claimed experience
- Career history
- Skill durations
This helps reduce the impact of synthetic or misleading candidate profiles.
Job Description ↓ Requirement Extraction ↓ Candidate Processing ↓ Semantic Scoring ↓ Behavioral Scoring ↓ Consistency Validation ↓ Composite Ranking ↓ Top Candidate Selection
TalentForge-AI-Ranker
│
├── rank.py
├── requirements.txt
├── submission_metadata.yaml
├── submission.csv
└── README.md
The generated submission contains:
- candidate_id
- rank
- score
- reasoning
- Transformer-based embeddings
- FAISS vector retrieval
- Learning-to-rank models
- LLM-generated explanations
- Active recruiter feedback loops
Purshottam Rakesh
India Runs 2026 – Data & AI Challenge