Niloofar Maani

Prepared a dataset derived from Tissue Micro Array slides for processing with PyTorch models. Developed a deep learning algorithm to predict PD-L1 immune scores using Hematoxylin and Eosin (H&E) stained slides.

Implemented an U-Net model for intra coding of fisheye video sequences. Implemented a self-driving car prototype using Raspberry Pi and OpenCV. This enabled live video processing for lane detection and maintaining proper direction control.

Implemented Acoustic Echo Cancellation method using adaptive filtering and NLMS algorithm to reduce the effect of additive noise. Implemented a framework based on a modern end-to-end Automatic Speech Recognition model called NeMo that supports evaluating the WER for processed speech utterances.

Responsible for programming various microcontrollers and developing online soft- ware. Developed web-based software to manage energy consumption of various devices in petrochemical plants. Developed and maintained an IoT system prototype for uploading temperature sensor data to a website in an industrial environment, utilizing Arduino and a wireless module (ESP8266).