app.py

import gradio as gr
from huggingface_hub import InferenceClient

data = '''
SARATH CHANDRA BANDREDDI
# PROFESSIONAL SUMMARY:
```
    Intern-level Python Developer focused on AI and Django. Skilled in Python, AI development, and building robust web 
    applications. Dedicated to continuous learning and staying updated with the latest industry trends and technologies. Eager to
    contribute to innovative software solutions and work collaboratively in a team environment.
```
# TECH STACK:
```
    ➢ Languages : Python, Java, JavaScript, C, R
    ➢ Scripting : Shell Scripting
    ➢ Markup Languages : HTML, CSS, Jinja Coding, UML
    ➢ Operating Systems : Linux, Windows
    ➢ IDE Tools : VS Code, RStudio, Pycharm
    ➢ Libraries/Frameworks : TensorFlow, Keras, LlamaIndex, Ollama, OpenCV, Sklearn, Numpy, Pandas, Django
    ➢ Tools / Platforms : Kaggle, Google-Colabs, Git, GitHub, AWS, Figma
    ➢ Databases : SQL, Oracle, My SQL
```
# EDUCATION PROFILE:
```
    ------------------------------------------------------------------------------------|
    | Course                     | Institution                      | CGPA | Duration   |
    |----------------------------|----------------------------------|------|------------|
    | B.Tech-CSE (AI & ML)       | VVIT - Vasireddy Venkatadri      | 8.59 | 2021-2025  |
    | 12th Class                 | Narayana Junior College          | 9.22 | 2020-2021  |
    | 10th Class                 | Narayana High School             | 9.7  | 2018-2019  |
    ------------------------------------------------------------------------------------|
```
# PROJECTS:
```
    ## FACE RECOGNITION VGGFace (LINK-"kaggle-notebook-link:https://www.kaggle.com/code/sarath02003/face-recognition-using-vggface2") -Individual project [Tensorflow | Keras | OpenCV |Django]:
    ```
        Developed a face recognition application utilizing the VGGFace architecture with TensorFlow, attaining exceptional accuracy on 
        a bespoke 29-class dataset. Launched the system through Hugging face space on an online server, showcasing advanced 
        capabilities in artificial neural networks and computer vision. Check the live application ("live-exegution-link:https://huggingface.co/spaces/Sarath0x8f/Celebrity_Recognition").
        Leveraged advanced facial recognition techniques using a custom dataset created to simulate real-world conditions and developed Celebrity Recognition Application. 
        Utilized data augmentation strategies, including shear, zoom, rotation, and brightness adjustments, to enhance model robustness. 
        Implemented transfer learning with the VGGFace model to accelerate training and improve accuracy to 98, incorporating custom fully connected layers and freezing certain layers for optimal performance. 
        Explored various optimizers (Adam, Adagrad, RMSprop) to optimize the model, demonstrating the critical role of optimizer selection in deep learning tasks. Integrated advanced training callbacks, such as EarlyStopping and ReduceLROnPlateau, 
        to prevent overfitting and ensure efficient training processes. Conducted thorough evaluations of the model on training and validation datasets, achieving significant improvements in loss and accuracy metrics. 
        Developed a deep understanding of data handling, model architecture customization, and training strategies, contributing to a more effective facial recognition model. 
        Looking forward to further advancements in the field of deep learning and facial recognition.
        - Mastered creating custom datasets using OpenCV, organizing training and test sets, and achieving 97.86% accuracy.
        - Trained deep learning models for image classification, leading to the publication of two papers in peer-reviewed journals 
        focusing on advancements in facial recognition technology.
    ```
    ## SCRAM (LINK-"Secure-Campus-Resource-and-Access-Management-link:https://github.com/21bq1a4210/E2E_Project/graphs/contributors") - Minor project [Gemini-pro (LLM) & Google API | Tensorflow | Keras | OpenCV | Django]:
    ```
        Engineered a robust, secure Django-based system for comprehensive college resource and access management, incorporating 
        advanced features such as user management, complaint registration, and a chatbot assistant. Seamlessly integrated Gemini-Pro 
        LLM with Google API and introduced face recognition technology for enhanced attendance tracking.
        Led the development of a robust and secure Django-based system designed to streamline college resource and access management as part of my End-to-End Project. The platform integrates several key functionalities such as user management, complaint registration, attendance management system, and a chatbot assistant.
        Here are my contribution to the project:
        - User Authentication & Security: Developed a comprehensive user authentication system using Django's default authentication and implemented 2-Step Verification (2SV) for password recovery, improving overall system security and reliability. Engineered a single-page application for password recovery with two-step verification, enhancing user convenience.
        - Face Recognition Attendance System: Created a one short face recognition model using FaceNet and MTCNN to manage attendance, with a unique feature allowing students to mark their attendance only once per day and within the campus premises. This innovation ensures strict attendance integrity and security.
        - Chatbot Integration: Built and integrated the AskVVIT chatbot to assist with college-related inquiries. Initially deployed with the Gemini Pro LLM and Google API, the chatbot provided an interactive platform for students and staff. Due to response time limitations (one response per minute), the model was later replaced with LLaMA 3.2:1B and also tried with LLaMA 3:latest, significantly enhancing response efficiency.
        - Backend & Django: Developed Django templates using Jinja and integrating frontend pages with backend functionality. Created models for user registration and attendence management system.

        This project not only enhanced resource management at the college but also introduced modern technologies such as face recognition and AI-driven chatbots, setting a foundation for future advancements in academic institution management systems. 
        • Devised robust user authentication and 2-FS password authentication, enhancing system security and reliability.
        • Led the project team, developing comprehensive Django templates, seamlessly integrating custom chatbot functionalities 
        using LLM and created a one short face recognition for attendance management system .
    ```
    ## Document and Data Query AI Agent - Individual project [Llama3 | Ollama | LlamaIndex]
    ```
        Created a sophisticated AI agent for document-based Q&A and CSV data visualization, utilizing Llama3 and Ollama for cuttingedge natural language processing. Integrated Llama Indexing for efficient data retrieval and robust document parsing within an 
        intuitive user interface.
        • Combined HuggingFace embedding generation and a resilient query retry mechanism, boosting data accuracy and system 
        reliability.
        • Demonstrated deep expertise in deep learning and real-time data processing, significantly improving system performance 
        and user experience.
    ```
    ## Personal Portfolio - Individual project [HTML | CSS | JS]:
    ```
        Designed and developed a responsive personal portfolio("LINK:https://21bq1a4210.github.io/MyPortfolio-/") website using HTML, CSS, and JavaScript, showcasing strong UI/UX principles. Deployed the website online using GitHub Pages, making it accessible for users worldwide. The website features multiple sections, including Home, About, Services, Portfolio, and Contact, providing a comprehensive overview of my skills and projects. Implemented interactive elements like a contact form integrated with EmailJS for seamless communication and added an AI chatbot, Tidio's Lyro, to enhance user engagement and provide additional information about me. Crafted unique visuals using AI-generated images to personalize the portfolio and demonstrate creativity. This project ignited my interest in web development, transforming initial reluctance into enthusiasm for building dynamic and interactive web experiences.
    ```
    ## OBJECT SEGMENTATION - Individual project [yoloV8, RoboFlow]:
    ```
        Built a versatile Python script using OpenCV and scikit image for image processing tasks.like object segmentation, using YoloV8, OpenCV, sklearn. Implemented functionalities like grayscale contrast nhancement, image similarity measurement (SSIM), and object segmentation YOLOv8 Demonstrated the script s capabilities through image processing, evaluation, and visualization. Planned to explore integration of advanced deep learning models for further functionalities.
        Here is the ("notebook link: Object_Segmentation_&_ComparisonObject_Segmentation_&_Comparison.ipynb:https://colab.research.google.com/drive/1cpNEx_u70I26Ex0alZZxoFa47p6fpoDz?usp=sharing")
    ```
```
#CERTIFICATION:
```
    ➢ Python Programming : Kaggle, SoloLearn, HackerRank, GUVI.
    ➢ Machine Learning : Coursera, Kaggle, Microsoft, IBM
    ➢ Deep Learning : Kaggle, NPTEL, OpenCVUniversity, IBM.
    ➢ Google Skill Boost(profile) : Introduction to Gen AI LP, Gemini for Google Cloud LP, Gen AI for Developers Learning Paths.
    ➢ Django : Microsoft
    ➢ AWS : AWS Academy Cloud Foundations, AWS Academy Cloud Architecting
    ➢ NPTEL (exam) : Java (ELITE), Data Science (ELITE + SILVER)
```
#ACHIEVEMENTS:
```
    Secure runner up in a Python Intramural technical fest at RVR&JC college (LINK)
```
# Contact me or Hire me:
```
    You can use below link to contace me.
    ## media:
    ```
        - LinkedIn:"https://www.linkedin.com/in/sarath-chandra-bandreddi-07393b1aa/"
        - MyPortfolio:"https://21bq1a4210.github.io/MyPortfolio-/"
        - 
    ```
```
'''

client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")


# Chatbot response function with integrated system message
def respond(
        message,
        history: list[tuple[str, str]],
        max_tokens=1024,
        temperature=0.5,
        top_p=0.95,
):
    # System message defining assistant behavior
    system_message = {
        "role": "system",
        "content": f"Act and chat as SARATH who is a professional fresher seeking a job and your name is SARATH."
                   f"Here is about you SARATH: data=```{data}```. You should answer questions based on this information only."
                   f'''Hire me or Contact me:
                       - LinkedIn:"https://www.linkedin.com/in/sarath-chandra-bandreddi-07393b1aa/"
                       - MyPortfolio:"https://21bq1a4210.github.io/MyPortfolio-/" '''
    }

    messages = [system_message]

    # Adding conversation history
    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    # Adding the current user input
    messages.append({"role": "user", "content": message})

    response = ""

    # Streaming the response from the API
    for message in client.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
    ):
        token = message.choices[0].delta.content
        response += token
        yield response

chatInterfaceCSS = '''
        width: 20%;  
        # height: auto;  
        border-radius: 75px;  
        box-shadow: 0 4px 12px rgba(0, 0, 0, 0.1);  
        padding: 10px; 
    '''

css = '''
    .contact-links {
        font-size: 1rem;  /* Slightly larger font */
        text-align: center;  /* Center-align the contact section */
        color: #0073e6;  /* Link color */
    }
    .contact-links a {
        color: #0073e6;  /* Ensure links are a nice blue */
        text-decoration: none;  /* Remove underline */
        padding: 0.5rem;  /* Add some padding around links */
        display: inline-block;
    }
    .contact-links a:hover {
        text-decoration: underline;  /* Underline on hover */
        color: red;
    }
'''

# Gradio interface with additional sliders for control
with gr.Blocks(theme=gr.themes.Soft(font=[gr.themes.GoogleFont("Roboto Mono")])) as demo:
    gr.Markdown('# Welcome to the DearHRSpeakWithMy2.0 🤖💬!')
    gr.Markdown(
        '''
            DearHRSpeakWithMy2.0 is a smart, AI-powered chatbot designed to act as a virtual introduction tool for job candidates during HR interviews. The bot is equipped to present comprehensive details about the candidate's skills, projects, and experience in a personalized and professional manner.
            Inspired by a real-world experience where the interviewer overlooked key aspects of the candidate’s expertise in AI and Machine Learning (ML), this project aims to ensure that important, job-relevant information is effectively communicated. DearHRSpeakWithMy2.0 helps avoid situations where an interviewer might focus on areas that don't align with the candidate's strengths or goals.
        '''
    )
    gr.ChatInterface(respond, css=chatInterfaceCSS)
    gr.Markdown(
        '''
                <h3>Contact Me:</h3><br>
                <a href="https://www.linkedin.com/in/sarath-chandra-bandreddi-07393b1aa/" target="_blank">LinkedIn</a> |
                <a href="https://21bq1a4210.github.io/MyPortfolio-/" target="_blank">My Portfolio</a> |
                <a href="mailto:[email protected]">Personal Email</a> |
                <a href="mailto:[email protected]">College Email</a> 
        '''
    )


if __name__ == "__main__":
    demo.launch(share=True)