Guneet Singh Bagga

Engineered and implemented a deep learning model, utilizing OpenCV, TensorFlow, and Keras, to enhance document clarity and readability by effectively reducing noise, glare, and distortions.

Presented project findings and technical approach to Mr. Prakash Velayudham, Founder & CEO of Codebuilders Pvt. Ltd., demonstrating practical application and industry relevance.

Achieved selection for publication as a research paper at the ICIOT’25 Conference, validating the model's innovation and scientific rigor.

Secured 3rd place at Project Expo 2025 for this innovative deep learning solution.

Performed comprehensive Exploratory Data Analysis (EDA) to identify critical features impacting crop yield, informing targeted agricultural strategies.

Developed and deployed Random Forest and Decision Tree models, achieving high accuracy in crop recommendation by integrating advanced feature engineering techniques.

Enhanced model robustness against diverse climate patterns through iterative feature engineering, ensuring reliable predictions under varying environmental conditions.

Secured publication as a research paper at the ICIOT’25 Conference, recognizing the project's significant contribution to agricultural science.

Applied LoRA (Low-Rank Adaptation) and PEFT to fine-tune LLaMA-2, achieving parameter-efficient model adaptation for specialized applications.

Constructed robust supervised fine-tuning pipelines leveraging Transformers, TRL SFTTrainer, and PyTorch, successfully customizing LLaMA-2 for specific custom datasets.

Implemented BitsAndBytes 4-bit quantization, significantly optimizing model training and inference to enable efficient utilization of limited GPU resources.

Developed core HTTP/1.1 functionalities, including precise request parsing, dynamic response generation, and effective client session management.

Architected a modular server design, significantly improving scalability and code maintainability for future enhancements.

Optimized server performance using Socket Programming and POSIX Threads, enabling efficient handling of multiple concurrent client connections.