Update app.py

app.py

@@ -1,139 +1,229 @@
-import random
-from typing import List, Dict
-import pandas as pd
 import gradio as gr
+import pandas as pd
+import numpy as np
 from sentence_transformers import SentenceTransformer
 from sklearn.metrics.pairwise import cosine_similarity
+import random
+from typing import List, Dict
+import os
 
-# ------------------------------
-# 1. Generate synthetic job database
-# ------------------------------
-def generate_job_database() -> List[Dict]:
-    job_templates = {
-        "Technology": [
-            {"title": "Software Engineer", "desc": "Design and develop software applications",
-             "skills": ["Python", "Java", "JavaScript", "Git", "Agile", "Problem Solving"]},
-            {"title": "Data Scientist", "desc": "Analyze complex data to extract business insights",
-             "skills": ["Python", "R", "Machine Learning", "SQL", "Statistics", "Pandas"]},
-        ],
-        "Healthcare": [
-            {"title": "Registered Nurse", "desc": "Provide patient care and medical support",
-             "skills": ["Patient Care", "Medical Knowledge", "CPR", "Communication", "Compassion"]},
-            {"title": "Medical Doctor", "desc": "Diagnose and treat medical conditions",
-             "skills": ["Medical Diagnosis", "Patient Care", "Clinical Skills", "Medical Ethics", "Communication"]},
-        ],
-        "Finance": [
-            {"title": "Financial Analyst", "desc": "Analyze financial data and market trends",
-             "skills": ["Financial Modeling", "Excel", "Data Analysis", "Financial Reporting", "Market Research"]},
-            {"title": "Accountant", "desc": "Manage financial records and tax preparation",
-             "skills": ["Accounting", "QuickBooks", "Tax Preparation", "Financial Reporting", "Attention to Detail"]},
-        ],
-        "Marketing": [
-            {"title": "Digital Marketing Manager", "desc": "Develop and execute digital marketing strategies",
-             "skills": ["Digital Marketing", "SEO", "Social Media", "Google Analytics", "Content Strategy"]},
-            {"title": "Content Marketing Specialist", "desc": "Create engaging content for marketing campaigns",
-             "skills": ["Content Creation", "SEO", "Social Media", "Writing", "Brand Management"]},
-        ],
-        "Education": [
-            {"title": "Elementary School Teacher", "desc": "Teach fundamental subjects to young students",
-             "skills": ["Teaching", "Classroom Management", "Curriculum Development", "Student Assessment", "Communication"]},
-            {"title": "High School Teacher", "desc": "Educate teenagers in specialized subjects",
-             "skills": ["Subject Expertise", "Teaching", "Classroom Management", "Lesson Planning", "Student Mentoring"]},
-        ],
-        "Sales": [
-            {"title": "Sales Representative", "desc": "Sell products and services to customers",
-             "skills": ["Sales", "Customer Relations", "Communication", "Negotiation", "Product Knowledge"]},
-            {"title": "Account Manager", "desc": "Manage relationships with key clients",
-             "skills": ["Account Management", "Client Relations", "Sales", "Communication", "Problem Solving"]},
-        ],
-        "Operations": [
-            {"title": "Operations Manager", "desc": "Oversee daily business operations",
-             "skills": ["Operations Management", "Process Improvement", "Team Leadership", "Budget Management", "Problem Solving"]},
-            {"title": "Project Manager", "desc": "Lead and coordinate project execution",
-             "skills": ["Project Management", "PMP", "Agile", "Risk Management", "Communication"]},
-        ],
-        "Creative": [
-            {"title": "Graphic Designer", "desc": "Create visual designs for various media",
-             "skills": ["Adobe Creative Suite", "Graphic Design", "Typography", "Brand Design", "Creativity"]},
-            {"title": "Content Writer", "desc": "Create written content for various platforms",
-             "skills": ["Writing", "Content Creation", "SEO", "Research", "Editing"]},
-        ]
-    }
-
-    experience_levels = ["Entry-level", "Mid-level", "Senior", "Lead/Principal"]
-
-    jobs = []
-    job_id = 1
-    categories = list(job_templates.keys())
-    jobs_per_category = 1000 // len(categories)
-    remaining_jobs = 1000 % len(categories)
-
-    for i, category in enumerate(categories):
-        templates = job_templates[category]
-        jobs_for_this_category = jobs_per_category + (1 if i < remaining_jobs else 0)
-
-        for j in range(jobs_for_this_category):
-            template = templates[j % len(templates)]
-            title_variations = [
-                template["title"],
-                f"Senior {template['title']}",
-                f"Junior {template['title']}",
-                f"Lead {template['title']}",
-                f"{template['title']} Specialist"
-            ]
-            title = title_variations[j % len(title_variations)]
-            exp_level = random.choice(experience_levels)
-            job = {
-                "id": job_id,
-                "title": title,
-                "description": template["desc"],
-                "requirements": template["skills"],
-                "experience_level": exp_level,
-                "category": category,
-                "location": random.choice([
-                    "Remote", "New York, NY", "San Francisco, CA", "Chicago, IL",
-                    "Austin, TX", "Seattle, WA", "Boston, MA", "Los Angeles, CA"
-                ])
-            }
-            jobs.append(job)
-            job_id += 1
-    return jobs
-
-# ------------------------------
-# 2. Prepare DataFrame and Embeddings
-# ------------------------------
-jobs = generate_job_database()
-df = pd.DataFrame(jobs)
-model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
-job_embeddings = model.encode(df['requirements'].apply(lambda x: " ".join(x)).tolist())
-
-# ------------------------------
-# 3. Job Recommendation Function
-# ------------------------------
-def recommend_jobs(user_skills, top_k=5):
-    user_embedding = model.encode([" ".join(user_skills)])
-    similarities = cosine_similarity(user_embedding, job_embeddings)[0]
-    top_indices = similarities.argsort()[-top_k:][::-1]
-    top_jobs = df.iloc[top_indices]
-    return top_jobs[['title', 'description', 'requirements', 'experience_level', 'category', 'location']]
-
-# ------------------------------
-# 4. Gradio Interface (v4.44.0 compatible)
-# ------------------------------
-def gradio_interface(skills_text):
-    user_skills = [skill.strip() for skill in skills_text.split(",")]
-    recommended = recommend_jobs(user_skills)
-    return recommended.to_dict(orient="records")
-
-with gr.Blocks() as app:
-    gr.Markdown("# AI Job Finder")
-    gr.Markdown("Enter your skills (comma-separated) to find the best matching jobs!")
-    
-    skills_input = gr.Textbox(lines=2, placeholder="Python, SQL, Machine Learning")
-    results_output = gr.Dataframe(headers=["Title", "Description", "Skills", "Experience", "Category", "Location"])
+class JobRecommendationSystem:
+    def __init__(self):
+        # Load a lightweight sentence transformer model for CPU
+        print("🤖 Loading AI model...")
+        self.model = SentenceTransformer('all-MiniLM-L6-v2')
+        
+        # Generate comprehensive job database
+        print("📊 Generating job database...")
+        self.jobs_data = self._generate_job_database()
+        
+        # Pre-compute job embeddings for efficiency
+        print("🧠 Computing job embeddings... This may take a moment.")
+        self.job_descriptions = [f"{job['title']} {job['description']} {' '.join(job['requirements'])}" 
+                                for job in self.jobs_data]
+        
+        try:
+            self.job_embeddings = self.model.encode(self.job_descriptions, show_progress_bar=False)
+            print(f"✅ Successfully loaded {len(self.jobs_data)} jobs!")
+        except Exception as e:
+            print(f"⚠️ Warning: Error computing embeddings: {e}")
+            # Fallback: compute embeddings in smaller batches
+            batch_size = 50
+            embeddings_list = []
+            for i in range(0, len(self.job_descriptions), batch_size):
+                batch = self.job_descriptions[i:i+batch_size]
+                batch_embeddings = self.model.encode(batch, show_progress_bar=False)
+                embeddings_list.append(batch_embeddings)
+            self.job_embeddings = np.vstack(embeddings_list)
+            print(f"✅ Successfully loaded {len(self.jobs_data)} jobs with batch processing!")
     
-    btn = gr.Button("Find Jobs")
-    btn.click(fn=gradio_interface, inputs=skills_input, outputs=results_output)
-
-if __name__ == "__main__":
-    app.launch()
+    def _generate_job_database(self) -> List[Dict]:
+        """Generate a comprehensive database of 1000 jobs across various industries"""
+        
+        # Job templates by category
+        job_templates = {
+            "Technology": [
+                {"title": "Software Engineer", "desc": "Design and develop software applications", 
+                 "skills": ["Python", "Java", "JavaScript", "Git", "Agile", "Problem Solving"]},
+                {"title": "Data Scientist", "desc": "Analyze complex data to extract business insights", 
+                 "skills": ["Python", "R", "Machine Learning", "SQL", "Statistics", "Pandas"]},
+                {"title": "DevOps Engineer", "desc": "Manage infrastructure and deployment pipelines", 
+                 "skills": ["AWS", "Docker", "Kubernetes", "Linux", "Python", "Terraform"]},
+                {"title": "Frontend Developer", "desc": "Create user interfaces and web experiences", 
+                 "skills": ["JavaScript", "React", "CSS", "HTML", "TypeScript", "Responsive Design"]},
+                {"title": "Backend Developer", "desc": "Build server-side applications and APIs", 
+                 "skills": ["Python", "Node.js", "Django", "PostgreSQL", "REST APIs", "MongoDB"]},
+                {"title": "Mobile Developer", "desc": "Develop applications for mobile platforms", 
+                 "skills": ["React Native", "Swift", "Kotlin", "Flutter", "iOS", "Android"]},
+                {"title": "QA Engineer", "desc": "Test software and ensure quality standards", 
+                 "skills": ["Selenium", "Test Automation", "Python", "Manual Testing", "JIRA"]},
+                {"title": "UI/UX Designer", "desc": "Design user interfaces and experiences", 
+                 "skills": ["Figma", "Adobe XD", "User Research", "Prototyping", "Design Systems"]},
+                {"title": "Machine Learning Engineer", "desc": "Deploy and maintain ML models in production", 
+                 "skills": ["Python", "TensorFlow", "PyTorch", "MLOps", "Docker", "Scikit-learn"]},
+                {"title": "Cybersecurity Analyst", "desc": "Monitor and protect against security threats", 
+                 "skills": ["Network Security", "Python", "SIEM", "Incident Response", "Penetration Testing"]},
+                {"title": "Database Administrator", "desc": "Manage and optimize database systems", 
+                 "skills": ["SQL", "MySQL", "PostgreSQL", "Database Design", "Performance Tuning"]},
+                {"title": "Cloud Architect", "desc": "Design scalable cloud infrastructure solutions", 
+                 "skills": ["AWS", "Azure", "GCP", "Cloud Architecture", "Microservices", "Serverless"]},
+                {"title": "Product Manager", "desc": "Define product strategy and roadmap", 
+                 "skills": ["Product Strategy", "Analytics", "Agile", "User Stories", "Market Research"]},
+                {"title": "Systems Administrator", "desc": "Maintain IT infrastructure and systems", 
+                 "skills": ["Linux", "Windows Server", "Network Administration", "Virtualization", "Shell Scripting"]},
+                {"title": "Full Stack Developer", "desc": "Work on both frontend and backend development", 
+                 "skills": ["JavaScript", "Python", "React", "Node.js", "SQL", "Git"]}
+            ],
+            "Healthcare": [
+                {"title": "Registered Nurse", "desc": "Provide patient care and medical support", 
+                 "skills": ["Patient Care", "Medical Knowledge", "CPR", "Communication", "Compassion"]},
+                {"title": "Medical Doctor", "desc": "Diagnose and treat medical conditions", 
+                 "skills": ["Medical Diagnosis", "Patient Care", "Clinical Skills", "Medical Ethics", "Communication"]},
+                {"title": "Physical Therapist", "desc": "Help patients recover from injuries", 
+                 "skills": ["Rehabilitation", "Exercise Therapy", "Patient Assessment", "Anatomy", "Communication"]},
+                {"title": "Medical Technologist", "desc": "Perform laboratory tests and analysis", 
+                 "skills": ["Laboratory Skills", "Medical Testing", "Quality Control", "Attention to Detail", "Safety Protocols"]},
+                {"title": "Pharmacist", "desc": "Dispense medications and provide drug information", 
+                 "skills": ["Pharmaceutical Knowledge", "Patient Counseling", "Drug Interactions", "Attention to Detail", "Regulatory Compliance"]},
+                {"title": "Medical Assistant", "desc": "Support healthcare providers with patient care", 
+                 "skills": ["Medical Procedures", "Patient Communication", "Medical Records", "Scheduling", "Basic Clinical Skills"]},
+                {"title": "Healthcare Administrator", "desc": "Manage healthcare facility operations", 
+                 "skills": ["Healthcare Management", "Budget Management", "Regulatory Compliance", "Leadership", "Strategic Planning"]},
+                {"title": "Radiologic Technologist", "desc": "Perform diagnostic imaging procedures", 
+                 "skills": ["Radiology", "Medical Imaging", "Patient Safety", "Equipment Operation", "Anatomy Knowledge"]}
+            ],
+            "Finance": [
+                {"title": "Financial Analyst", "desc": "Analyze financial data and market trends", 
+                 "skills": ["Financial Modeling", "Excel", "Data Analysis", "Financial Reporting", "Market Research"]},
+                {"title": "Investment Banker", "desc": "Provide financial advisory services", 
+                 "skills": ["Financial Analysis", "Valuation", "Excel", "Presentation Skills", "Client Relations"]},
+                {"title": "Accountant", "desc": "Manage financial records and tax preparation", 
+                 "skills": ["Accounting", "QuickBooks", "Tax Preparation", "Financial Reporting", "Attention to Detail"]},
+                {"title": "Risk Analyst", "desc": "Assess and manage financial risks", 
+                 "skills": ["Risk Assessment", "Statistical Analysis", "Financial Modeling", "Regulatory Knowledge", "Problem Solving"]},
+                {"title": "Portfolio Manager", "desc": "Manage investment portfolios", 
+                 "skills": ["Investment Strategy", "Portfolio Analysis", "Market Research", "Risk Management", "Client Communication"]},
+                {"title": "Credit Analyst", "desc": "Evaluate creditworthiness of borrowers", 
+                 "skills": ["Credit Analysis", "Financial Modeling", "Risk Assessment", "Banking Regulations", "Excel"]},
+                {"title": "Insurance Underwriter", "desc": "Evaluate insurance applications and risks", 
+                 "skills": ["Risk Assessment", "Insurance Knowledge", "Data Analysis", "Decision Making", "Attention to Detail"]},
+                {"title": "Tax Consultant", "desc": "Provide tax planning and preparation services", 
+                 "skills": ["Tax Law", "Tax Preparation", "Client Consultation", "Attention to Detail", "Regulatory Compliance"]}
+            ],
+            "Marketing": [
+                {"title": "Digital Marketing Manager", "desc": "Develop and execute digital marketing strategies", 
+                 "skills": ["Digital Marketing", "SEO", "Social Media", "Google Analytics", "Content Strategy"]},
+                {"title": "Content Marketing Specialist", "desc": "Create engaging content for marketing campaigns", 
+                 "skills": ["Content Creation", "SEO", "Social Media", "Writing", "Brand Management"]},
+                {"title": "Social Media Manager", "desc": "Manage social media presence and engagement", 
+                 "skills": ["Social Media", "Content Creation", "Community Management", "Analytics", "Brand Voice"]},
+                {"title": "SEO Specialist", "desc": "Optimize websites for search engine visibility", 
+                 "skills": ["SEO", "Google Analytics", "Content Optimization", "Keyword Research", "Technical SEO"]},
+                {"title": "Marketing Analyst", "desc": "Analyze marketing data and campaign performance", 
+                 "skills": ["Data Analysis", "Google Analytics", "Marketing Metrics", "Excel", "Reporting"]},
+                {"title": "Brand Manager", "desc": "Develop and maintain brand identity and strategy", 
+                 "skills": ["Brand Strategy", "Marketing", "Creative Direction", "Market Research", "Communication"]},
+                {"title": "Email Marketing Specialist", "desc": "Create and manage email marketing campaigns", 
+                 "skills": ["Email Marketing", "Automation", "A/B Testing", "Analytics", "Copywriting"]},
+                {"title": "PPC Specialist", "desc": "Manage pay-per-click advertising campaigns", 
+                 "skills": ["Google Ads", "PPC", "Campaign Management", "Analytics", "Budget Management"]}
+            ],
+            "Education": [
+                {"title": "Elementary School Teacher", "desc": "Teach fundamental subjects to young students", 
+                 "skills": ["Teaching", "Classroom Management", "Curriculum Development", "Student Assessment", "Communication"]},
+                {"title": "High School Teacher", "desc": "Educate teenagers in specialized subjects", 
+                 "skills": ["Subject Expertise", "Teaching", "Classroom Management", "Lesson Planning", "Student Mentoring"]},
+                {"title": "Special Education Teacher", "desc": "Work with students with special needs", 
+                 "skills": ["Special Education", "IEP Development", "Adaptive Teaching", "Patience", "Communication"]},
+                {"title": "School Counselor", "desc": "Provide academic and personal guidance to students", 
+                 "skills": ["Counseling", "Student Support", "Academic Planning", "Crisis Intervention", "Communication"]},
+                {"title": "Principal", "desc": "Lead and manage school operations", 
+                 "skills": ["Educational Leadership", "Staff Management", "Budget Management", "Policy Development", "Communication"]},
+                {"title": "Librarian", "desc": "Manage library resources and assist patrons", 
+                 "skills": ["Information Management", "Research Skills", "Library Systems", "Customer Service", "Organization"]},
+                {"title": "Educational Technology Specialist", "desc": "Integrate technology into educational environments", 
+                 "skills": ["Educational Technology", "Training", "Technical Support", "Curriculum Integration", "Problem Solving"]},
+                {"title": "Instructional Designer", "desc": "Design educational programs and materials", 
+                 "skills": ["Instructional Design", "Curriculum Development", "Learning Theory", "Educational Technology", "Project Management"]}
+            ],
+            "Sales": [
+                {"title": "Sales Representative", "desc": "Sell products and services to customers", 
+                 "skills": ["Sales", "Customer Relations", "Communication", "Negotiation", "Product Knowledge"]},
+                {"title": "Account Manager", "desc": "Manage relationships with key clients", 
+                 "skills": ["Account Management", "Client Relations", "Sales", "Communication", "Problem Solving"]},
+                {"title": "Sales Manager", "desc": "Lead sales teams and develop strategies", 
+                 "skills": ["Sales Management", "Team Leadership", "Strategic Planning", "Performance Management", "Communication"]},
+                {"title": "Business Development Manager", "desc": "Identify and develop new business opportunities", 
+                 "skills": ["Business Development", "Sales", "Market Research", "Relationship Building", "Strategic Thinking"]},
+                {"title": "Inside Sales Representative", "desc": "Conduct sales activities remotely", 
+                 "skills": ["Phone Sales", "CRM", "Lead Generation", "Communication", "Product Knowledge"]},
+                {"title": "Real Estate Agent", "desc": "Help clients buy and sell properties", 
+                 "skills": ["Real Estate", "Customer Service", "Negotiation", "Market Knowledge", "Communication"]},
+                {"title": "Retail Sales Associate", "desc": "Assist customers in retail environments", 
+                 "skills": ["Customer Service", "Sales", "Product Knowledge", "Cash Handling", "Communication"]},
+                {"title": "Territory Sales Manager", "desc": "Manage sales activities in specific geographic areas", 
+                 "skills": ["Territory Management", "Sales", "Travel", "Customer Relations", "Strategic Planning"]}
+            ],
+            "Operations": [
+                {"title": "Operations Manager", "desc": "Oversee daily business operations", 
+                 "skills": ["Operations Management", "Process Improvement", "Team Leadership", "Budget Management", "Problem Solving"]},
+                {"title": "Supply Chain Manager", "desc": "Manage supply chain and logistics operations", 
+                 "skills": ["Supply Chain", "Logistics", "Vendor Management", "Process Optimization", "Analytics"]},
+                {"title": "Project Manager", "desc": "Lead and coordinate project execution", 
+                 "skills": ["Project Management", "PMP", "Agile", "Risk Management", "Communication"]},
+                {"title": "Quality Assurance Manager", "desc": "Ensure product and service quality standards", 
+                 "skills": ["Quality Management", "Process Improvement", "ISO Standards", "Auditing", "Problem Solving"]},
+                {"title": "Logistics Coordinator", "desc": "Coordinate transportation and distribution", 
+                 "skills": ["Logistics", "Transportation", "Inventory Management", "Coordination", "Problem Solving"]},
+                {"title": "Business Analyst", "desc": "Analyze business processes and requirements", 
+                 "skills": ["Business Analysis", "Requirements Gathering", "Process Mapping", "Data Analysis", "Communication"]},
+                {"title": "Warehouse Manager", "desc": "Manage warehouse operations and staff", 
+                 "skills": ["Warehouse Management", "Inventory Control", "Team Leadership", "Safety Management", "Logistics"]},
+                {"title": "Production Manager", "desc": "Oversee manufacturing and production processes", 
+                 "skills": ["Production Management", "Manufacturing", "Process Optimization", "Quality Control", "Team Leadership"]}
+            ],
+            "Creative": [
+                {"title": "Graphic Designer", "desc": "Create visual designs for various media", 
+                 "skills": ["Adobe Creative Suite", "Graphic Design", "Typography", "Brand Design", "Creativity"]},
+                {"title": "Web Designer", "desc": "Design websites and digital interfaces", 
+                 "skills": ["Web Design", "HTML", "CSS", "Adobe Creative Suite", "User Experience"]},
+                {"title": "Video Editor", "desc": "Edit and produce video content", 
+                 "skills": ["Video Editing", "Adobe Premiere", "Final Cut Pro", "Motion Graphics", "Storytelling"]},
+                {"title": "Content Writer", "desc": "Create written content for various platforms", 
+                 "skills": ["Writing", "Content Creation", "SEO", "Research", "Editing"]},
+                {"title": "Art Director", "desc": "Lead creative vision for projects", 
+                 "skills": ["Creative Direction", "Team Leadership", "Brand Strategy", "Visual Design", "Project Management"]},
+                {"title": "Photographer", "desc": "Capture and edit professional photographs", 
+                 "skills": ["Photography", "Photo Editing", "Adobe Lightroom", "Photoshop", "Composition"]},
+                {"title": "Animator", "desc": "Create animated content and motion graphics", 
+                 "skills": ["Animation", "After Effects", "3D Animation", "Motion Graphics", "Storytelling"]},
+                {"title": "Copywriter", "desc": "Write compelling marketing and advertising copy", 
+                 "skills": ["Copywriting", "Marketing", "Brand Voice", "Persuasive Writing", "Creative Thinking"]}
+            ]
+        }
+        
+        # Experience levels and salary ranges
+        experience_levels = ["Entry-level", "Mid-level", "Senior", "Lead/Principal"]
+        
+        salary_ranges = {
+            "Entry-level": ["$35k-$50k", "$40k-$55k", "$45k-$60k", "$50k-$65k"],
+            "Mid-level": ["$55k-$75k", "$60k-$80k", "$65k-$85k", "$70k-$90k"],
+            "Senior": ["$80k-$110k", "$90k-$120k", "$100k-$130k", "$110k-$140k"],
+            "Lead/Principal": ["$120k-$150k", "$130k-$160k", "$140k-$170k", "$150k-$180k"]
+        }
+        
+        # Additional skills by category
+        additional_skills = {
+            "Technology": ["Problem Solving", "Debugging", "Code Review", "System Design", "API Development", "Database Design"],
+            "Healthcare": ["HIPAA Compliance", "Medical Ethics", "Electronic Health Records", "Patient Safety", "Clinical Documentation"],
+            "Finance": ["Financial Regulations", "Risk Management", "Excel Advanced", "Bloomberg Terminal", "Financial Compliance"],
+            "Marketing": ["Brand Strategy", "Customer Acquisition", "Marketing Automation", "CRM", "Market Analysis"],
+            "Education": ["Student Engagement", "Assessment", "Educational Psychology", "Classroom Technology", "Differentiated Instruction"],
+            "Sales": ["CRM Systems", "Lead Qualification", "Sales Forecasting", "Territory Planning", "Customer Retention"],
+            "Operations": ["Lean Six Sigma", "Process Documentation", "KPI Management", "Vendor Relations", "Cost Reduction"],
+            "Creative": ["Brand Identity", "Design Thinking", "Client Presentation", "Creative Strategy", "Visual Communication"]
+        }
+        
+        # Generate 1000 jobs
+        jobs = []
+        job_id = 1