README.md

---
title: SHL Assessment Recommendation Engine
emoji: 🎯
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 4.44.0
app_file: app.py
pinned: false
license: mit
short_description: AI-powered SHL assessment recommendations using 2-Stage RAG
---

# 🎯 SHL Assessment Recommendation Engine

AI-powered recommendations for SHL assessments using a novel **2-Stage RAG Pipeline**.

## 🏗️ Architecture

```
Query → SBERT Encoder → FAISS Index → Top-K → Cross-Encoder Reranker → Results
```

| Component | Model | Purpose |
|-----------|-------|---------|
| **Retriever** | all-MiniLM-L6-v2 | Fast semantic embedding |
| **Index** | FAISS IndexFlatIP | Efficient similarity search |
| **Reranker** | ms-marco-MiniLM-L-6-v2 | Accurate pairwise scoring |
| **Scoring** | Hybrid | 30% retrieval + 70% reranking |

## 📊 Performance

| Metric | Score | Interpretation |
|--------|-------|----------------|
| **P@1** | 0.90 | 90% top-1 accuracy |
| **MRR** | 0.95 | First relevant at rank ~1 |
| **NDCG@5** | 0.944 | Near-optimal ranking |

## 🚀 Usage

1. Enter a job description or role requirements
2. Click "Get Recommendations"
3. View matching SHL assessments with scores

## 📚 Dataset

74 SHL assessments from the [SHL Product Catalog](https://www.shl.com/products/product-catalog/) including:
- Job Focused Assessments
- Cognitive Assessments
- Personality Assessments
- Skills & Simulations

---

Built for SHL Research Internship Assessment • December 2024