app.py

import gradio as gr
from sentence_transformers import SentenceTransformer, util

# Load embedding model (small + free CPU friendly)
model = SentenceTransformer("all-MiniLM-L6-v2")

# Example dataset of jobs & required skills
jobs = [
    {"title": "Data Scientist", "skills": "python, machine learning, data analysis, statistics"},
    {"title": "Web Developer", "skills": "html, css, javascript, react"},
    {"title": "Graphic Designer", "skills": "photoshop, illustrator, creativity, ui design"},
    {"title": "Digital Marketer", "skills": "seo, social media, content creation, advertising"},
    {"title": "Project Manager", "skills": "planning, communication, leadership, risk management"},
    {"title": "Cybersecurity Analyst", "skills": "network security, risk assessment, ethical hacking"},
    {"title": "Data Engineer", "skills": "sql, big data, etl, data pipelines"},
]

# Precompute job embeddings
job_embeddings = model.encode([job["skills"] for job in jobs], convert_to_tensor=True)

def recommend_jobs(user_skills):
    # Encode user input
    user_embedding = model.encode(user_skills, convert_to_tensor=True)
    # Compute cosine similarity
    scores = util.cos_sim(user_embedding, job_embeddings)[0]
    # Sort top 3
    top_results = scores.topk(3)
    
    recommendations = []
    for idx in top_results.indices:
        job = jobs[int(idx)]
        recommendations.append(f"{job['title']} — Skills: {job['skills']}")
    return "\n".join(recommendations)

# Gradio UI
iface = gr.Interface(
    fn=recommend_jobs,
    inputs=gr.Textbox(lines=3, placeholder="Enter your skills (comma separated)..."),
    outputs="text",
    title="AI Job Recommender",
    description="Type in your skills and get 3 job suggestions that match your profile."
)

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