Work Experience
Wielabs Software Development Company, Junior software developer (Full time)
04/2024 – present | Hyderabad, India
  • •Worked on training and fine-tuning deep learning models for real-world AI/​ML use cases, including Audiobook generation and LLM-powered applications.
  • •Built end-to-end systems with FastAPI for backend development, integrated LLMs into frontend workflows, and managed databases for scalable deployment.
    • Eizen.AI, Machine learning engineer (Intern)
    08/2023 – 10/2023 | Hyderabad, India
  • •Worked on multi-object class detection using pre-trained YOLO v4.
  • •Built featured APIs using the Python FastAPI and moviePY library.
    • Suntek Corp Solutions Pvt Ltd⁠, DSA technical teaching assistant (Intern)
    03/2023 – 07/2023 | Hyderabad, India
  • •Collaborated with faculty to refine the curriculum for data structures, incorporating real-world problems.
  • •Ensuring alignment with industry standards, and updates positively impacted 150+ enrolled students.
    • Education
    Maturi Venkata Subba Rao Engineering College - B.Tech(CSE)
    2020 – 2024

    CGPA : 8.18

    Kakatiya Junior College - Intermediate
    2018 – 2020

    Percentage : 95.5%

    Vignan High School - SSC
    2008 – 2018

    GPA : 8.70

    Projects
  • •Developed a modular video automation pipeline using FastAPI backend, enabling clips processing, including transcription, remixing, B-roll injection, music overlay, and aesthetic presets.

  • •LLM-based keyword extraction for B-rolls and music to generate shortform, styled content automatically with real-time feedback via Socket.IO
  • •Built a context-aware RAG chatbot using OpenAI’s GPT models and vector search, enabling real-time responses grounded in internal documents and knowledge bases.
    • Receipt text classifier using YOLOv5 (Python), July 2024
  • •The receipt text classifier using YOLOv5 is about classifying the information of receipts by extracting from the receipts using OCR.
  • • Using extracted classified information, further data analysis like increase in GST%, monthly expenses, and visualizing the data.
    • Publications

    Sarcasm classification in Telugu faces NLP challenges due to limited data and model constraints. We use FastText and IndicFT embeddings with classifiers like SVM, decision tree, KNN, random forest, and Naive Bayes. To handle class imbalance, we apply undersampling, oversampling, ensembling, and anomaly detection techniques. SVM performs best with up to 0.80 validation accuracy.

    Skills
    Machine learning
    Python programming
    NoSQL
    Artificial intelligence