DEVANSH CHOUDHARY

Detail-oriented Computer Science graduate with experience in full-stack application development and backend services. Skilled in Java, Spring Boot, .NET, and MERN stack, with expertise in RESTful APIs and scalable application design. Capable of delivering reliable software solutions and collaborating in team-oriented environments.

• Microsoft Azure Fundamentals (AI-900)
• Complete Python Bootcamp – Udemy
• Power Platform Fundamentals – Microsoft
• Computer System Security – IIT Kanpur

• LLM for Educational Content Generation
  Built and fine‑tuned a decoder‑only transformer‑based LLM to generate quizzes and summaries.
  Integrated OpenAI pre‑trained weights and custom data pipelines, fine‑tuned with education‑specific datasets to align with curriculum standards.
• Retrieval Augmented Generation (RAG) System
  Implemented RAG using Llama3, LangChain, and ChromaDB for document question‑answering.
  Designed vector database retrieval to enhance contextual accuracy before generation.
• DoS/DDoS Detection & Mitigation for IoT
  Developed a protocol‑based deep intrusion detection (PB‑DID) system achieving 96.3% accuracy in IoT traffic classification.
  Applied Python, TensorFlow, and Keras with advanced techniques to address imbalanced data and overfitting, strengthening IoT security.
• Lung Disease Classification using CNNs
  GitHub Repository
  Built a CNN‑based model on X‑ray images, achieving 91.45% accuracy with DenseNet201 and VGG16.
  Delivered deep learning insights for enhanced clinical diagnostics using TensorFlow and Keras.
• OpenCV Motion Detection
  Implemented motion detection using background subtraction and erosion to isolate moving objects and reduce noise.
  Optimized for security, surveillance, and real‑time analytics applications.

• Date of Birth: 11/22/01
• Gender: Male
• Nationality: Indian

• English
• Hindi