Resume Optimizer

from llama_index.core import SimpleDirectoryReader, Settings, VectorStoreIndex
from llama_index.embeddings.nebius import NebiusEmbedding
from llama_index.llms.nebius import NebiusLLM

from llama_index.core import SimpleDirectoryReader
import tempfile
import shutil

def load_resume(uploaded_file):
    """Load and process PDF resume."""
    # Create temporary directory
    temp_dir = tempfile.mkdtemp()
    
    # Save uploaded PDF
    file_path = os.path.join(temp_dir, uploaded_file.name)
    with open(file_path, "wb") as f:
        f.write(uploaded_file.getbuffer())
    
    # Load with LlamaIndex
    documents = SimpleDirectoryReader(temp_dir).load_data()
    
    return documents, temp_dir

from llama_index.core import Settings
from llama_index.embeddings.nebius import NebiusEmbedding
from llama_index.llms.nebius import NebiusLLM
import os

# Initialize Nebius LLM
llm = NebiusLLM(
    model="Qwen/Qwen3-235B-A22B",  # Or "deepseek-ai/DeepSeek-V3"
    api_key=os.getenv("NEBIUS_API_KEY")
)

# Initialize Nebius embeddings
embed_model = NebiusEmbedding(
    model_name="BAAI/bge-en-icl",
    api_key=os.getenv("NEBIUS_API_KEY")
)

# Configure LlamaIndex settings
Settings.llm = llm
Settings.embed_model = embed_model

from llama_index.core import VectorStoreIndex

def run_rag_completion(
    documents,
    query_text: str,
    job_title: str,
    job_description: str,
    embedding_model: str = "BAAI/bge-en-icl",
    generative_model: str = "Qwen/Qwen3-235B-A22B"
) -> str:
    """Run RAG completion for resume optimization."""
    # Configure models
    llm = NebiusLLM(
        model=generative_model,
        api_key=os.getenv("NEBIUS_API_KEY")
    )
    
    embed_model = NebiusEmbedding(
        model_name=embedding_model,
        api_key=os.getenv("NEBIUS_API_KEY")
    )
    
    Settings.llm = llm
    Settings.embed_model = embed_model
    
    # Step 1: Analyze the resume
    analysis_prompt = f"""
    Analyze this resume in detail. Focus on:
    1. Key skills and expertise
    2. Professional experience and achievements
    3. Education and certifications
    4. Notable projects or accomplishments
    5. Career progression and gaps
    
    Provide a concise analysis in bullet points.
    """
    
    index = VectorStoreIndex.from_documents(documents)
    resume_analysis = index.as_query_engine(similarity_top_k=5).query(analysis_prompt)
    
    # Step 2: Generate optimization suggestions
    optimization_prompt = f"""
    Based on the resume analysis and job requirements, provide specific, actionable improvements.
    
    Resume Analysis:
    {resume_analysis}
    
    Job Title: {job_title}
    Job Description: {job_description}
    
    Optimization Request: {query_text}
    
    Provide a direct, structured response in this exact format:

    ## Key Findings
    • [2-3 bullet points highlighting main alignment and gaps]

    ## Specific Improvements
    • [3-5 bullet points with concrete suggestions]
    • Each bullet should start with a strong action verb
    • Include specific examples where possible

    ## Action Items
    • [2-3 specific, immediate steps to take]
    • Each item should be clear and implementable

    Keep all points concise and actionable.
    """
    
    optimization_suggestions = index.as_query_engine(similarity_top_k=5).query(optimization_prompt)
    
    return str(optimization_suggestions)

optimization_prompts = {
    "ATS Keyword Optimizer": 
        "Identify and optimize ATS keywords. Focus on exact matches and semantic variations from the job description.",
    
    "Experience Section Enhancer": 
        "Enhance experience section to align with job requirements. Focus on quantifiable achievements.",
    
    "Skills Hierarchy Creator": 
        "Organize skills based on job requirements. Identify gaps and development opportunities.",
    
    "Professional Summary Crafter": 
        "Create a targeted professional summary highlighting relevant experience and skills.",
    
    "Education Optimizer": 
        "Optimize education section to emphasize relevant qualifications for this position.",
    
    "Technical Skills Showcase": 
        "Organize technical skills based on job requirements. Highlight key competencies.",
    
    "Career Gap Framing": 
        "Address career gaps professionally. Focus on growth and relevant experience."
}

import streamlit as st
from PyPDF2 import PdfReader
import base64

st.set_page_config(page_title="Resume Optimizer", layout="wide")
st.title("📝 Resume Optimizer")
st.caption("Powered by Nebius AI")

# Sidebar: Configuration and resume upload
with st.sidebar:
    st.image("./Nebius.png", width=150)
    
    # Model selection
    generative_model = st.selectbox(
        "Generative Model",
        ["Qwen/Qwen3-235B-A22B", "deepseek-ai/DeepSeek-V3"],
        index=0
    )
    
    st.divider()
    
    # Resume upload
    st.subheader("Upload Resume")
    uploaded_file = st.file_uploader(
        "Choose your resume (PDF)",
        type="pdf",
        accept_multiple_files=False
    )
    
    if uploaded_file:
        # Process resume
        documents, temp_dir = load_resume(uploaded_file)
        st.session_state.docs_loaded = True
        st.session_state.documents = documents
        st.success("✓ Resume loaded successfully")
        
        # PDF Preview
        st.subheader("Resume Preview")
        base64_pdf = base64.b64encode(uploaded_file.getvalue()).decode('utf-8')
        pdf_display = f'<iframe src="data:application/pdf;base64,{base64_pdf}" width="100%" height="500" type="application/pdf"></iframe>'
        st.markdown(pdf_display, unsafe_allow_html=True)

# Main area: Job information and optimization
col1, col2 = st.columns([1, 1])

with col1:
    st.subheader("Job Information")
    job_title = st.text_input("Job Title")
    job_description = st.text_area("Job Description", height=200)
    
    st.subheader("Optimization Options")
    optimization_type = st.selectbox(
        "Select Optimization Type",
        [
            "ATS Keyword Optimizer",
            "Experience Section Enhancer",
            "Skills Hierarchy Creator",
            "Professional Summary Crafter",
            "Education Optimizer",
            "Technical Skills Showcase",
            "Career Gap Framing"
        ]
    )
    
    if st.button("Optimize Resume"):
        if not st.session_state.get('docs_loaded', False):
            st.error("Please upload your resume first")
        elif not job_title or not job_description:
            st.error("Please provide both job title and description")
        else:
            with st.spinner("Analyzing and generating suggestions..."):
                response = run_rag_completion(
                    st.session_state.documents,
                    optimization_prompts[optimization_type],
                    job_title,
                    job_description,
                    "BAAI/bge-en-icl",
                    generative_model
                )
                # Remove think tags
                response = response.replace("<think>", "").replace("</think>", "")
                st.session_state.messages.append({"role": "assistant", "content": response})

with col2:
    st.subheader("Optimization Results")
    if "messages" in st.session_state:
        for message in st.session_state.messages:
            st.markdown(message["content"])

# This happens automatically in LlamaIndex:
index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine(
    similarity_top_k=5  # Retrieve top 5 most similar chunks
)
response = query_engine.query(prompt)

git clone https://github.com/Arindam200/awesome-ai-apps.git
cd rag_apps/resume_optimizer
uv sync

NEBIUS_API_KEY=your_api_key_here

streamlit run main.py