T

Thirupugal S

About

Detail

Thanjavur, Tamil Nadu, India

Timeline

work
Job
school
Education
folder
Project

Résumé

Jobs verified_user 0% verified
  • I
    Data Annotator
    Intellect Design Arena (Infocareer Private Limited)
    Aug 2025 - Current (10 months)
    Performed PDF and image annotation for AI model training, creating precise bounding boxes and labels for object detection and text extraction tasks, supporting the AI team. Executed data normalization and preprocessing to ensure high-quality, consistent datasets for machine learning, enhancing model performance. Collaborated with data science and AI teams to validate annotations, improving model accuracy for document processing and computer vision applications.
  • VDart
    AI & ML Intern
    VDart
    Jan 2025 - May 2025 (5 months)
    Smart Interview System for Recruitment and Skill Assessment. Developed an AI-driven interview system, automating aptitude, Q&A, and HR levels with Google Meet integration via Pandas/smtplib. Implemented YOLOv5 for person detection and Faster R-CNN for mobile detection using PyTorch/OpenCV for real-time proctoring. Utilized LLaMA2 (via Ollama) for dynamic question generation; evaluated answers with TF-IDF, Sentence Transformers, and VADER Sentiment analysis. Created a Plotly dashboard for result filtering, visualization, and billing, Excel downloads with secure admin/candidate login, saving 90% of money, time, and manpower. Optimized detection with asyncio for CPU/GPU efficiency; stored results in Excel/JSON.
Education verified_user 0% verified
  • S
    M.Sc Data Science
    SASTRA Deemed to be University
    Aug 2023 - May 2025 (1 year 10 months)
    CGPA 8.3/10
  • A
    B.Sc Mathematics
    A.V.V.M Sri Pushpam College
    Aug 2020 - May 2023 (2 years 10 months)
    CGPA 8.72/10
Projects (professional or personal) verified_user 0% verified
  • P
    PDF Chatbot Using Llama 3 and RAG Model
    Aug 2024 - Dec 2024 (5 months)
    PDF Upload and Processing Enables seamless uploading of PDF documents by users. Utilizes PyPDFLoader for precise extraction and analysis of document content. RAG with Llama 3 Employs Retrieval-Augmented Generation using the Llama 3 model. Generates accurate, context-aware responses derived from uploaded document data. Vector Embeddings Employs advanced embeddings for efficient document indexing. Facilitates rapid and relevant information retrieval to enhance response quality. Interactive Chat Interface Features an intuitive web interface supporting dynamic user conversations with real-time chat history visualization.
  • A
    Advanced Speech Emotion Recognition Using Combined Feature Extraction and Deep Learning Models
    Dec 2023 - Aug 2024 (9 months)
    Enhanced Speech Emotion Recognition (SER) by integrating MFCC numerical features and Mel-spectrogram images. Applied PCA and trained a Convolution Neural Network (CNN) on numerical features; used pretrained VGG for spectrogram image analysis. Built a hybrid model with improved emotion recognition accuracy, achieving better performance on the TESS dataset. Applicable in virtual assistants, mental health monitoring, and interactive AI systems.