About

Humayun Irshad is an accomplished Engineering Manager of Machine Learning at Pano, a wildfire situational awareness intelligence platform that combines disparate data feeds including proprietary mountaintop cameras and geostationary satellites, with deep learning AI to provide actionable insights. With over a decade of experience in Machine Learning and Computer Vision, Humayun is a specialist in deep learning and active learning with domain expertise in Climate Tech, Facial Identity, Medical Imaging, and self-driving vehicles. Humayun has delivered computer vision products and APIs for use cases including image recognition and classification, object detection and segmentation in 2D and 3D images/videos - application ranging from climate, satellite, medical, facial identity, self-driving vehicles, retail, and fashion. He has applied and trained others to apply deep learning algorithms (CNN, RCNN, LSTM, Transformers), image processing methods, feature extraction techniques, pattern recognition, feature selection/regularization, boosting methodologies, generalized regression models, data visualization, analytic approximation theory, enumerative and graphical combinatorics, advanced implementations of operator algebras, probability theory, and statistics. In the past, Humayun has developed machine learning and deep learning models that include region of interests detection and classification, and nuclei and gland detection, segmentation, and classification in 2D and 3D histopathological images (H&E stained, IHC stained, and fluorescence images). He has also developed a CAD System for Breast screening using Mammographic and MRI images included image enhancement and standardization, lump/tumor/mass detection, segmentation, and classification, breast tissue characterization, and eventually computing BIRAD Score. At Pano, Humayun is developing a ML-based early smoke/fire detection, more specifically wildfire. He is designing a scalable and compatible architecture for different sets of sensors (camera, Satellite, IOTs, etc.), cloud and edge-based solutions. He is also improving the wildfire detection algorithms by optimizing both detection speed and accuracy (reduce False Positives & False Negatives) and designing and implementing a multi-year roadmap for continuous improvement in AI-based smoke detection solutions and infrastructure. Additionally, he is building a team of ML scientists, engineering, and DevOps. Humayun has worked with Mesmer, a cloud-based Robotics Process Automation (RPA) tool for mobile app developers, and Synthesis AI, where he developed a pipeline for generation and testing of synthetic images (faces, retail products, etc.). He has a Ph.D