Update app.py

app.py

@@ -25,10 +25,10 @@ from pydantic import BaseModel, Field, confloat
 # )
 from langchain_community.llms import HuggingFaceHub
 #HuggingFaceH4/zephyr-7b-beta"
-llm = HuggingFaceHub(
-    repo_id="mistralai/Mistral-7B-Instruct-v0.2",
-    task="text-generation",
-)
+# llm = HuggingFaceHub(
+#     repo_id="mistralai/Mistral-7B-Instruct-v0.2",
+#     task="text-generation",
+# )
 
 class SkillScore(BaseModel):
     skill_name: str = Field(description="Name of the skill being scored")
@@ -255,7 +255,7 @@ resume_analyzer = Agent(
                                 providing actionable suggestions for improvement. Your recommendations always
                                 focus on both human readability and ATS compatibility.""",
             verbose=True,
-            llm = llm,
+            #llm = llm,  # Commenting out llm here
             # knowledge_sources=[pdf_source],
         )
 job_analyzer = Agent(
@@ -268,7 +268,7 @@ job_analyzer = Agent(
                 and soft skills requirements, and can evaluate experience levels accurately.""",
             verbose=True,
             tools=[ScrapeWebsiteTool()],
-            llm = llm,
+            #llm = llm, # Commenting out llm here
 
             # knowledge_sources=[pdf_source],
 )
@@ -282,7 +282,7 @@ company_researcher = Agent(
                 candidates for interviews. """,
             tools=[SerperDevTool()],
             verbose=True,
-            llm = llm,
+            #llm = llm,  # Commenting out llm here
 
             # knowledge_sources=[pdf_source],
 
@@ -296,7 +296,7 @@ resume_writer = Agent (
                     beautifully formatted, ATS-friendly documents that maintain professionalism
                     while showcasing candidate strengths effectively.""",
                 verbose=True,
-                llm = llm,
+                #llm = llm,  # Commenting out llm here
 
         )
 report_generator = Agent(
@@ -309,49 +309,42 @@ report_generator = Agent(
                     into clear, actionable insights with proper markdown formatting, emojis, and
                     visual elements that make information both appealing and easily digestible.""",
                 verbose=True,
-                llm = llm,
+                #llm = llm, # Commenting out llm here
 
         )
 
 analyze_job_task = Task(
     description="""Analyze the {job_url} description and score the candidate's fit based on their resume.
                     Output will be saved as structured JSON data.
-
                     1. Extract Requirements:
                     - Technical skills (required vs nice-to-have)
                     - Soft skills
                     - Experience levels
                     - Education requirements
                     - Industry knowledge
-
                     2. Score Technical Skills (35% of total):
                     - For each required skill:
                         * Match Level (0-1): How well does candidate's experience match?
                         * Years Experience: Compare to required years
                         * Context Score: How relevant is their usage of the skill?
                     - Calculate weighted average based on skill importance
-
                     3. Score Soft Skills (20% of total):
                     - Identify soft skills from resume
                     - Compare against job requirements
                     - Consider context and demonstration of skills
-
                     4. Score Experience (25% of total):
                     - Years of relevant experience
                     - Role similarity
                     - Industry relevance
                     - Project scope and complexity
-
                     5. Score Education (10% of total):
                     - Degree level match
                     - Field of study relevance
                     - Additional certifications
-
                     6. Score Industry Knowledge (10% of total):
                     - Years in similar industry
                     - Domain expertise
                     - Industry-specific achievements
-
                     7. Calculate Overall Score:
                     - Weighted average of all components
                     - Identify key strengths and gaps
@@ -367,19 +360,16 @@ optimize_resume_task = Task(
     description= """
                 Review the provided resume against the job analysis and create structured optimization suggestions.
                 Output will be saved as structured JSON data.
-
                 1. Content Analysis:
                 - Compare resume content with job requirements
                 - Identify missing keywords and skills
                 - Analyze achievement descriptions
                 - Check for ATS compatibility
-
                 2. Structure Review:
                 - Evaluate section organization
                 - Check formatting consistency
                 - Assess information hierarchy
                 - Verify contact details
-
                 3. Generate Suggestions:
                 - Content improvements with before/after examples
                 - Skills to highlight based on job match
@@ -403,7 +393,6 @@ research_company_task = Task(
                 - Culture and values
                 - Market position
                 - Growth trajectory
-
                 2. Interview Preparation:
                 - Common interview questions
                 - Company-specific topics
@@ -422,19 +411,16 @@ generate_resume_task = Task(
                             Using the optimization suggestions and job analysis from previous steps,
                             create a polished resume in markdown format.
                             Do not add markdown code blocks like '```'.
-
                             1. Content Integration:
                             - Incorporate optimization suggestions
                             - Add missing keywords and skills
                             - Enhance achievement descriptions
                             - Ensure ATS compatibility
-
                             2. Formatting:
                             - Use proper markdown headers (#, ##, ###)
                             - Apply consistent styling
                             - Create clear section hierarchy
                             - Use bullet points effectively
-
                             3. Documentation:
                             - Track changes made
                             - Note preserved elements
@@ -453,39 +439,32 @@ generate_report_task = Task(
             description="""
                         Create an executive summary report using data from previous steps.
                         Format in markdown without code blocks '```'.
-
                         1. Data Integration:
                         - Job analysis and scores
                         - Resume optimization details
                         - Company research insights
                         - Final resume changes
-
                         2. Report Sections:
                         ## Executive Summary
                         - Overall match score and quick wins
                         - Key strengths and improvement areas
                         - Action items priority list
-
                         ## Job Fit Analysis
                         - Detailed score breakdown
                         - Skills match assessment
                         - Experience alignment
-
                         ## Optimization Overview
                         - Key resume improvements
                         - ATS optimization results
                         - Impact metrics
-
                         ## Company Insights
                         - Culture fit analysis
                         - Interview preparation tips
                         - Key talking points
-
                         ## Next Steps
                         - Prioritized action items
                         - Skill development plan
                         - Application strategy
-
                         3. Formatting:
                         - Use proper markdown headers
                         - Include relevant emojis
@@ -504,7 +483,7 @@ from crewai import Crew
 # Function to run CrewAI with dynamic inputs
 def run_crew(api_key: str, job_url: str, company_name: str, resume_pdf_path: str):
     if not api_key:
-        return "⚠️ Please provide a valid OpenAI API Key."
+        return "⚠️ Please provide a valid HUGGINGFACEHUB_API_TOKEN API Key."
 
     os.environ["HUGGINGFACEHUB_API_TOKEN"] = f"{api_key}"  # Set API key securely
     # os.environ["HUGGINGFACEHUB_API_TOKEN"] = userdata.get('HF_TOKEN')
@@ -515,6 +494,19 @@ def run_crew(api_key: str, job_url: str, company_name: str, resume_pdf_path: str
     file_paths=[resume_pdf_path])
     # Initialize CrewAI with resume knowledge
     print(resume_pdf_path)
+    # Initialize HuggingFaceHub with API token
+    llm = HuggingFaceHub(
+        repo_id="mistralai/Mistral-7B-Instruct-v0.2",
+        task="text-generation",
+        huggingfacehub_api_token=api_key,
+    )
+    # Create agents with the initialized llm
+    resume_analyzer.llm = llm
+    job_analyzer.llm = llm
+    company_researcher.llm = llm
+    resume_writer.llm = llm
+    report_generator.llm = llm
+
     crew = Crew(
     agents=[resume_analyzer, job_analyzer, company_researcher, resume_writer, report_generator],
     tasks=[analyze_job_task, optimize_resume_task, research_company_task, generate_resume_task, generate_report_task ],
@@ -529,21 +521,6 @@ def run_crew(api_key: str, job_url: str, company_name: str, resume_pdf_path: str
 
     # Run AI pipeline
     result_extractor = crew.kickoff(inputs=inputs)
-    # Ensure output exists before writing files
-    # optimized_resume_text = result_extractor.get('optimized_resume') if isinstance(result_extractor, dict) and 'optimized_resume' in result_extractor else "No resume data generated."
-    # final_report_text = result_extractor.get('final_report') if isinstance(result_extractor, dict) and 'final_report' in result_extractor else "No report generated."
-
-    # # Save outputs as markdown files
-    # optimized_resume_path = os.path.join(OUTPUT_DIR, "optimized_resume.md")
-    # final_report_path = os.path.join(OUTPUT_DIR, "final_report.md")
-
-    # with open(optimized_resume_path, "w", encoding="utf-8") as resume_file:
-    #     resume_file.write(f"# Optimized Resume\n\n{optimized_resume_text}\n")
-
-    # with open(final_report_path, "w", encoding="utf-8") as report_file:
-    #     report_file.write(f"# AI Resume Optimization Report\n\n{final_report_text}\n")
-
-    # return "✅ Resume processed successfully! Download the files below.", optimized_resume_path, final_report_path
 
     return result_extractor
 
@@ -590,7 +567,7 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     gr.Markdown("# 🤖 AI Resume Optimizer 🚀")
     gr.Markdown("Get Complete analysis of your resume")
 
-    api_key = gr.Textbox(label="🔑 Open AI API Key", placeholder="Enter your OpenAI API Key securely")
+    api_key = gr.Textbox(label="🔑 HUGGINGFACEHUB_API_TOKEN Key", placeholder="Enter your HUGGINGFACEHUB_API_TOKEN securely")
     job_url = gr.Textbox(label="🔗 Job URL", placeholder="Paste job listing URL here")
     company_name = gr.Textbox(label="🏢 Company Name", placeholder="Enter company name")
     resume_pdf = gr.File(label="📄 Upload Resume (PDF Only)")
@@ -605,4 +582,4 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
 
 # Launch the UI
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()