At Hurix Digital, I trained AI models with audio and video data, working with large language models to improve model performance. I also integrated AI-driven features into a web-based digital publishing platform.
For my Soil Erosion & Land Degradation Detection project, I built a deep neural network image classifier that reached 87% accuracy on a four-class imbalanced dataset. The work was published at the SRM 4th International Conference and recognized by the American Book of Records at Star Summit 2024.
I also built SmartBreeze, an IoT climate-control system using DHT22 sensors and an NVIDIA Jetson Nano. It automatically regulated fan speed within ±0.5°C, while my Python scripts processed sensor data in real time.
