Sejal jadhav

​Developed and optimized 15+ deep learning models, including Vision Transformers and UNet architectures for video analysis, improving quality metrics by 5% through architectural refinements. Conducted 80+ MLflow experiments with GPU profiling across video codecs and optimized model architectures. Designed a scalable ML pipeline processing 2TB+ video data with cross-functional teams, achieving 92% accuracy.

Delivered a time series solar energy forecasting model achieving 8.7% MAPE and 21% MAE improvement, with automated scheduled retraining pipelines. Architected and deployed a production-grade RAG chatbot on Azure OpenAI Service, fine-tuning GPT, LLaMA, and Mistral for domain-specific knowledge retrieval and respons. Designed end-to-end NLP model workflows covering tokenization, prompt engineering, hyperparameter tuning, and cloud deployment.

Engineered end-to-end Computer Vision pipelines leveraging YOLO and Faster R-CNN for automated detection and classification of structural elements across rural retail environments. Achieved 25% improvement in inventory placement and product accessibility through deep learning-based object detection, enabling data-driven planogram optimization. Applied spatial analysis algorithms to redesign store layouts, delivering 15 - 20% gains in operational efficiency across multiple locations

Specialization in Machine Learning & Artificial Intelligence