At BharatSoftwareSolutions, I built and deployed a CNN wheat disease detection model using TensorFlow and OpenCV, achieving 99%+ validation accuracy across five disease classes. I also integrated the model into a Flutter app for real-time crop diagnosis.
In my LLM Fine-Tuning Pipeline project, I fine-tuned open-source models with Hugging Face Transformers and LoRA/QLoRA, then built a RAG pipeline deployed via FastAPI. My projects also include a generative AI pipeline for medical imaging and drug discovery using Conditional GANs and a ResNet-18 encoder.
