app.py

import pandas as pd
import numpy as np
import faiss
from sentence_transformers import SentenceTransformer
import gradio as gr

# ---------- Load CSV ----------
df = pd.read_csv("shl_products.csv")

# ---------- Description Helper ----------
def create_description(row):
    return f"{row['Pre-packaged Job Solutions']} is a {row['Test Type']} test. " \
           f"Remote Testing support: {row['Remote Testing']}. Adaptive/IRT: {row['Adaptive/IRT']}."

df["description"] = df.apply(create_description, axis=1)

# ---------- Sentence Embeddings ----------
model = SentenceTransformer("all-MiniLM-L6-v2")
df["embedding"] = df["description"].apply(lambda x: model.encode(str(x)))

# ---------- Build FAISS Index ----------
embedding_matrix = np.stack(df["embedding"].values)
dimension = embedding_matrix.shape[1]
index = faiss.IndexFlatL2(dimension)
index.add(embedding_matrix)

# ---------- Retrieval Function ----------
def recommend_assessments(query, k=10):
    query_embedding = model.encode(query)
    D, I = index.search(np.array([query_embedding]), k)
    results = df.iloc[I[0]][[
        "Pre-packaged Job Solutions",
        "Remote Testing",
        "Adaptive/IRT",
        "Test Type"
    ]]
    return results.reset_index(drop=True)

# ---------- Gradio UI ----------
interface = gr.Interface(
    fn=recommend_assessments,
    inputs=gr.Textbox(lines=3, label="Job Description / Hiring Need"),
    outputs=gr.Dataframe(type="pandas"),
    title="SHL Assessment Recommender",
    description="Enter a natural language hiring query and get relevant SHL assessments.",
    examples=[
        ["Looking for a cognitive test for engineers that supports remote testing."],
        ["Assessment for sales roles with adaptive questions and remote option."],
        ["Test for developers that evaluates collaboration and reasoning in under 40 minutes."]
    ]
)


if __name__ == "__main__":
    interface.launch()