import streamlit as st
from PIL import Image
from streamlit_lottie import st_lottie
import json
import os
import glob
from streamlit_option_menu import option_menu
from projects import display_projects

#setting layout to wide
st.set_page_config(layout="wide")
# Load CSS for styling with a minimalist grey background
with open("style.css") as f:
    css_content = f.read()
    css_content += '''
    body {
        background-color: #f0f2f6;
    }
    '''
    st.markdown('<style>{}</style>'.format(css_content), unsafe_allow_html=True)

def load_lottiefile(filepath: str):
    with open(filepath, "r") as file:
        return json.load(file)

def display_header():
    st.write('''
    # Nihar Palem
    ##### 
    ''')
    # Assuming you have a Lottie animation to display
    lottie_animation = load_lottiefile("bio.json")
    st_lottie(lottie_animation, height=300, key="header_animation")

def display_summary():
    #st.markdown('## Summary', unsafe_allow_html=True)

    st.markdown("""
Hello! I'm **Sai Nihar Reddy Palem**, an Applied AI Engineer, Data Scientist, and AI Researcher based in San Jose, California. Originally from Hyderabad, India, I've embarked on a transformative journey from Electrical Engineering to becoming a passionate AI researcher exploring the frontiers of artificial intelligence.

My path began with a solid engineering foundation, evolved through diverse industry experiences across data engineering and analytics, and reached new heights with my **Master's degree in Applied Data Science** from San Jose State University. Over the past 2+ years, I've immersed myself in the cutting-edge world of multimodal AI, Large Language Model fine-tuning, and multi-agent architectures, consistently pushing the boundaries between theoretical research and practical implementation. Currently, I'm working with Google on bettering their multimodal capabilities, contributing to the advancement of state-of-the-art AI systems.

What drives me is the exciting challenge of systematically translating breakthrough research papers into production-ready solutions that create real-world impact. From achieving 12% performance improvements through advanced LLM optimization techniques to actively participating and learning from technical events like hackathons where I've built innovative multi-agent systems in just 5 hours (most recent), my journey reflects a deep commitment to both research excellence and practical innovation. I've contributed to open-source projects that have garnered 10,000+ community interactions, developed comprehensive evaluation frameworks for state-of-the-art models like Gemini 3.0, and created an applications that democratize AI technology for businesses and individuals alike.

**What You'll Find in This Portfolio**: Education, Work Experience, Projects, Skills, Research Notes, Social Media, Open Source Applications, Awards

""")


def display_education():
    st.markdown('## Education')
    st.write("""
    - **Masters In Data Analytics**, *San Jose State University*, USA (2023-2024)
      - Courses: Data Mining, Deep Learning, Big Data Technologies, Data Visualization, Machine Learning, Database Management Systems
      - Achievements: 
        - A Grade in Deep Learning
    - **Bachelor of Technology (B.Tech) in Electrical and Electronics Engineering (EEE)**, *Sreenidhi Institute of Science and Technology (SNIST)*, Hyderabad (2015-2019)
      - Activities: 
        - Memeber of the Robotics Club:, built line follower and theft-alert detection bots.
        - Member of the college cricket team; Runner up in City-level tournament
    """)

def display_work_experience():
    st.markdown('## Work Experience')
    st.write("""
    
  **Turing, San Jose, CA, USA**
    March 2024 - Present
    - **Data Scientist & Applied AI Engineer**
        - Collaborated with AI engineers, product teams, researchers, and Google DeepMind team to integrate LLM evaluation systems into production workflows using PyTorch and distributed computing
        - Engineered comprehensive evaluation benchmarks for Gemini 3.0 by analyzing reasoning loss patterns and image loss patterns in state-of-the-art Vision-Language Models (VLMs) including o3 and Gemini 2.5 Pro, developing custom datasets across multiple domains (mathematics, finance, chemistry, biology) spanning educational levels from high-school through PhD with statistical validation methods
        - Implemented advanced LLM fine-tuning strategies for Qwen model including Parameter-Efficient Fine-Tuning (PEFT) with LoRA and 2-stage whole model training on multi-GPU clusters, achieving 12% performance improvement across 15+ categories
        - Developed "auto hinter" system to improve LLM reasoning, guiding models towards correct answers based on question complexity, resulting in 8% performance increment on PhD-level questions
        - Built "auto rater" system to assess responses from leading models like Gemini 2.5 Pro and o3 custom builds, scoring across four key dimensions: completeness, coherence, clarity, correctness, style and formatting
       
    **San Jose State University, San Jose, CA, USA**
    August 2024 - December 2024
    - **Teaching Assistant**
        - Mentored 80+ graduate students on data modeling projects, providing feedback on technical documentation
        - Reviewed and debugged student data pipelines, offering solutions for data analysis and ML model challenges
        - Improved student performance, with 75% of mentored students achieving an 'A' grade
        -Conducted weekly office hours to assist students with complex data science concepts and project implementations
    
    **Bharat Electronics Limited, Hyderabad, India**
    February 2021 - March 2022
    - **Data Analyst**
      - Optimized SQL queries for sales and payroll databases using indexes and CTEs, reducing execution times by 40%
      - Developed and maintained 20+ Tableau dashboards, reducing production costs by 15% and improving sales forecasts by 10%
      - Implemented automated billing checks using SQL procedures, reducing financial discrepancies by 30%
      - Optimized ETL pipelines with staging tables and data quality checks, increasing ingestion efficiency by 25%
      
    **Technical Writer**
    2023-Present
        - Embarked on a new journey in 2023 as a technical writer, sharing insights and developments in data science and data engineering with a growing audience.
        - Authored numerous articles that explore complex topics in an accessible and informative manner, focusing on AI, data science, machine learning and data engineering.
        - This new habit aims to educate and inspire, bridging the gap between technical expertise and practical application in the modern data landscape.
        - Find my work on [Medium](https://medium.com/@nihar-palem) and [Substack](https://niharpalem.substack.com/publish/posts).
    """)
def display_skills():
    st.title('Skills')
    
    # Define tab titles
    tab_titles = [
        "Programming & Core",
        "AI & ML",
        "Data Engineering",
        "Data Architecture",
        "Visualization",
        "Specialized Systems",
        "Multimodal AI",
        "LLM & Advanced AI"
    ]
    
    # Create tabs
    tabs = st.tabs(tab_titles)
    
    # Programming & Core Technologies
    with tabs[0]:
        st.subheader("Programming & Core Technologies")
        st.markdown("""
        - **Programming Languages**:
            - Python (Advanced)
            - SQL (Advanced)
            - Shell Scripting
            
        - **Database Systems**: 
            - Relational: MySQL, PostgreSQL
            - NoSQL: MongoDB
            - Data Warehouses: Snowflake, Redshift
            - Vector Databases: FAISS, Pinecone
            
        - **Development Tools**:
            - Version Control: Git, GitHub
            - Containerization: Docker
            - Orchestration: Kubernetes (Basic)
            - IDE: VS Code, PyCharm
            - Microsoft Office Suite
            
        - **Frameworks & Libraries**:
            - LangChain
            - Hugging Face (Transformers, Diffusers)
            - Scikit-Learn, Pandas, NumPy
            - Apache Spark
        """)
    
    # AI & Machine Learning
    with tabs[1]:
        st.subheader("AI & Machine Learning")
        st.markdown("""
        - **Machine Learning Frameworks**: 
            - PyTorch (Advanced, PyTorch Distributed, DDP)
            - TensorFlow
            - Scikit-Learn
            - XGBoost, Random Forest, AdaBoost
            
        - **Deep Learning**: 
            - Vision Transformers (ViT)
            - Vision Language Models
            - Large Language Models
            - Sentecne Transformers
            - Diffusion Models
            - ResNet Architectures
            - Neural Networks
            - BiLSTM
            
        - **Distributed Training**:
            - Multi-GPU Clusters (16+ GPUs)
            - PyTorch DDP (Distributed Data Parallel)
            - DeepSpeed
            - Megatron
            - CUDA Acceleration
            - FlashAttention
            
        - **Computer Vision**: 
            - MediaPipe
            - OpenCV
            - Image Processing Pipelines
            - Satellite Imagery Analysis
            
        - **Model Optimization**:
            - Model Compression (Quantization, Distillation)
            - Performance Optimization
            - CUDA Programming
            - Parallel Processing
        """)
    
    # Data Engineering & Cloud
    with tabs[2]:
        st.subheader("Data Engineering & Cloud")
        st.markdown("""
        - **Cloud Platforms**: 
            - AWS (Certified - Lambda, S3, Glue, EC2, Redshift)
            - Google Cloud Platform (GCP)
            - Cloud Architecture Design
            
        - **Big Data Technologies**: 
            - Apache Spark (PySpark)
            - Apache Airflow
            - BigQuery
            - Hadoop Ecosystem
            
        - **Data Pipeline Tools**: 
            - ETL/ELT Pipeline Design
            - Workflow Orchestration
            - Concurrent Processing
            - Real-time Data Streaming
            - ThreadPoolExecutor Optimization
            
        - **Infrastructure**:
            - CI/CD Pipelines (GitHub Actions)
            - Infrastructure as Code
            - Kubernetes Basics
            - Production Monitoring
            - Distributed Training Clusters
        """)
    
    # Data Architecture & Analytics
    with tabs[3]:
        st.subheader("Data Architecture & Analytics")
        st.markdown("""
        - **Data Modeling**: 
            - OLAP/OLTP Systems
            - Star/Snowflake Schema
            - Data Normalization
            - Database Optimization
            
        - **Analytics Techniques**:
            - Streaming Analytics
            - Batch Processing
            - Time Series Analysis
            - Statistical Analysis
            - A/B Testing
            - Hypothesis Testing
            
        - **Data Processing**: 
            - Pandas, NumPy
            - Data Wrangling
            - Feature Engineering
            - Data Quality Assurance
            - Data Quality Management
            
        - **Performance Optimization**:
            - Query Optimization
            - Indexing Strategies
            - Caching Mechanisms
            - SQL Performance Tuning
        """)
    
    # Visualization & Deployment
    with tabs[4]:
        st.subheader("Visualization & Tools")
        st.markdown("""
        - **Business Intelligence**: 
            - Tableau
            - Power BI
            - Dashboard Design
            - KPI Monitoring
            
        - **Technical Visualization**: 
            - Plotly
            - Seaborn
            - Matplotlib
            - Interactive Charts
            
        - **Deployment & Interface**: 
            - Streamlit
            - Web Development
            - Hugging Face Spaces
            
        - **Collaboration Tools**: 
            - JIRA
            - Notion
            - Git Workflow
            - Agile Methodologies
        """)
    
    # Specialized Systems
    with tabs[5]:
        st.subheader("Specialized Systems")
        st.markdown("""
        - **Recommender Systems**: 
            - Hybrid Filtering Techniques
            - Content-Based Filtering
            - Collaborative Filtering
            - Matrix Factorization (SVD)
            
        - **Ensemble Methods**: 
            - Multi-model Consensus Systems
            - Classifier Combinations
            - Voting Systems
            - Stacking Implementations
            
        - **Performance Optimization**: 
            - CUDA Acceleration
            - Parallel Processing
            - Resource Management
            - Scalability Design
            
        - **Custom Solutions**:
            - Natural Language Processing
            - Computer Vision Systems
            - Time Series Forecasting
            - Anomaly Detection
            - Real-time Web Scraping
            - Automated Data Quality Checks
        """)
    
    # Multimodal AI
    with tabs[6]:
        st.subheader("Multimodal AI")
        st.markdown("""
        - **Vision-Language Models**: 
            - Qwen-VL
            - Gemini Multimodal
            - Vision-Language Understanding
            - Cross-modal Fine-tuning
            - Multimodal Evaluation
            
        - **Visual AI**: 
            - Visual Question Answering (VQA)
            - Vision Transformers (ViT)
            - Stable Diffusion XL
            - Generative AI (Vision)
            - Image-Text Alignment
            
        - **Multi-Agent Systems**: 
            - Multi-Agent Multimodal Workflows
            - Strategic Agent Architecture
            - Visual Agent Integration
            - QA Agent Implementation
            
        - **Evaluation & Testing**:
            - Multimodal Benchmarking
            - Cross-modal Bias Detection
            - Performance Optimization
            - Adversarial Testing
            - Statistical Validation Methods
        """)
    
    # LLM & Advanced AI
    with tabs[7]:
        st.subheader("LLM & Advanced AI")
        st.markdown("""
        - **Large Language Models**: 
            - Fine-tuning (PEFT, LoRA, QLoRA)
            - 2-Stage Training
            - VLLM/LMMs
            - Qwen, LLaMA (Llama-3.1-8B), GPT Integration
            
        - **Advanced Techniques**: 
            - Prompt Engineering (Advanced, Context Injection)
            - RAG (Retrieval-Augmented Generation)
            - LLM Evaluation Benchmarking
            - LLM-as-judge
            - Auto Hinter Systems
            
        - **Production AI Systems**: 
            - Multi-Agent Systems
            - API Integration
            - Performance Optimization
            - Tenstorrent Hardware Utilization
            - MLOps
            
        - **Specialized Applications**:
            - Semantic Job Matching
            - Resume Generation
            - Marketing Campaign Automation
            - Infrastructure Change Detection
            - Exercise Pose Correction
            
        - **AI Testing & Validation**:
            - Unit/Integration Testing for AI
            - Offline Evaluation Frameworks
            - Model Validation
            - ROC Curve Analysis
            - RMSE Validation
            - Bias Mitigation
        """)

def display_articles():
    """Display articles from HTML files in the articles directory"""
    st.markdown('## Articles')
    
    # Define the articles directory path
    articles_dir = "articles"  # You can change this path as needed
    
    # Check if articles directory exists
    if not os.path.exists(articles_dir):
        st.warning(f"Articles directory '{articles_dir}' not found. Please create the directory and add your HTML files.")
        st.info("Create an 'articles' folder in your project directory and add your HTML files there.")
        return
    
    # Get all HTML files from the articles directory
    html_files = glob.glob(os.path.join(articles_dir, "*.html"))
    
    if not html_files:
        st.info("No HTML articles found in the articles directory. Add some .html files to get started!")
        return
    
    # Sort files by name for consistent ordering
    html_files.sort()
    
    st.markdown("Click on any article below to view:")
    st.markdown("")  # Add some space
    
    # Display each article as a clean clickable card
    for i, html_file in enumerate(html_files):
        # Extract filename without path and extension, format it nicely
        file_name = os.path.splitext(os.path.basename(html_file))[0]
        display_name = file_name.replace('_', ' ').replace('-', ' ').title()
        
        # Get file size
        try:
            file_size = os.path.getsize(html_file)
            size_kb = round(file_size / 1024, 1)
            size_text = f"{size_kb} KB"
        except:
            size_text = "- KB"
        
        # Create a clean card-like button
        article_clicked = st.button(
            f"📄 {display_name} ({size_text})",
            key=f"article_{i}",
            use_container_width=True
        )
        
        if article_clicked:
            try:
                with open(html_file, 'r', encoding='utf-8') as file:
                    html_content = file.read()
                
                # Display article content inline
                st.markdown("---")
                st.markdown(f"### {display_name}")
                
                # Add download option
                col1, col2 = st.columns([1, 4])
                with col1:
                    st.download_button(
                        label="⬇️ Download",
                        data=html_content,
                        file_name=os.path.basename(html_file),
                        mime="text/html",
                        key=f"download_clicked_{i}"
                    )
                with col2:
                    if st.button("❌ Close Article", key=f"close_{i}"):
                        st.rerun()
                
                # Display the HTML content
                st.components.v1.html(html_content, height=600, scrolling=True)
                
            except Exception as e:
                st.error(f"Error loading article: {str(e)}")
                
        # Add small spacing between articles
        st.markdown("")

def display_apps():
    st.markdown('## Apps')
    st.markdown("""
- [CNN arch](https://cnn-arch.streamlit.app/)
""")
    st.markdown("""
- [TuNNe](https://huggingface.co/spaces/Niharmahesh/TuNNe)
""")
    
def display_certifications():
    st.markdown('## Certifications')
    
    certifications = [
        {"title": "Python for Data Science and Machine Learning Bootcamp", "issuer": "Udemy", "date": "2023", "skills": "Python, Data Science, Machine Learning"},
        {"title": "HackerRank SQL (Basic)", "issuer": "HackerRank", "date": "2023", "skills": "SQL, Database Management"},
        {"title": "AWS Cloud Practitioner", "issuer": "Udemy", "date": "2023", "skills": "Cloud Computing, AWS Services"},
        {"title": "AWS Certified Cloud Practitioner", "issuer": "Amazon Web Services", "date": "2023", "skills": "Cloud Architecture, AWS Best Practices"}
    ]
    
    for cert in certifications:
        with st.expander(cert["title"]):
            st.write(f"**Issuer:** {cert['issuer']}")
            st.write(f"**Date:** {cert['date']}")
            st.write(f"**Skills:** {cert['skills']}")

def display_social_media():
    st.markdown('## Social Media')
    st.markdown("""
- [LinkedIn](https://www.linkedin.com/in/sai-nihar-1b955a183/)
- [GitHub](https://github.com/niharpalem)
- [Medium](https://medium.com/@nihar-palem)
- [Twitter](https://twitter.com/niharpalem_2497)
- [Email](mailto:sainiharreddy.palem@sjsu.edu)
""")

# Updated menu with articles section
menu_items_with_icons = {
    "🎓": display_education,
    "💼": display_work_experience,
    "📁": display_projects,
    "🛠️": display_skills,
    "📝": display_articles,  # New articles section
    "🌐": display_social_media,
    "🏆": display_certifications,
    "📱": display_apps
}

def main():
    # Initialize session state for selected function
    if 'selected_function' not in st.session_state:
        st.session_state.selected_function = None  # Default to None to not display any section initially

    # Display the header with your name and Lottie animation first
    display_header()
    
    # Display the summary section immediately after the header
    display_summary()

    # Create a row of buttons for each icon in the menu
    cols = st.columns(len(menu_items_with_icons))
    for col, (icon, func) in zip(cols, menu_items_with_icons.items()):
        if col.button(icon):
            # Update the session state to the selected function
            st.session_state.selected_function = func

    # If a function has been selected, call it
    if st.session_state.selected_function is not None:
        st.session_state.selected_function()

if __name__ == "__main__":
    main()