Hammad Ahmad Usmani

Analyzing the Existing Undergraduate Engineering Leadership Skills

Dr. Hamed M. Almalki, Dr. Luis Rabelo, Charles Davis, Hammad Usmani, Dr. Debra Hollister, Dr. Alfonso Sarmiento

A Deep Recurrent Neural Network to Forecast the Intensity and Trajectory of Atlantic Tropical Storms (2019)

Hammad Usmani, Georgia Institute of Technology, Atlanta, GA

This study presents a bidirectional deep recurrent neural network (BDRNN) utilizing LSTM cells to forecast Atlantic storm trajectories and intensity, outperforming statistical baselines like OCD5. Developed with HURDAT2 data, the BDRNN offers timely and precise emergency planning, highlighting the importance of advanced forecasting models in storm preparedness.

Global Synthetic Weather Radar in AWS GovCloud for the U.S. Air Force (2020)

Mark S. Veillette, Haig Iskenderian, Patrick M. Lamey, Christopher J. Mattioli, Ashish Banerjee, Mark Worris, Alexander B. Proschitsky, Richard F. Ferris, Artyom Manwelyan, Shibi Rajagopalan, Hammad Usmani, Thomas E. Coe, Jennifer E. Luce, Blaine A. Esgar

The U.S. Air Force, in collaboration with MIT Lincoln Laboratory, is advancing a machine learning tool for generating global radar-like mosaics for flight operations. Using data from lightning, the GALWEM model, and weather satellite images, a convolutional neural network creates global synthetic weather radar mosaics. Transitioned to the AWS GovCloud, it facilitates real-time evaluation and aids Air Force decision systems. The project entails capability development, cloud integration, user feedback, and ensuring sustainable machine learning practices.

A Deep Neural Network to Globally Forecast the Track and Intensity of Tropical Cyclones (2020)

Hammad Usmani, Aadil Habibi, Daanish Habibi

As tropical cyclones intensify with global warming, this study harnesses machine learning to predict their tracks and intensities. Utilizing the IBTrACS database and NCEP/NCAR Surface Temperature imagery, a deep neural network combining recurrent and convolutional layers is developed. An accompanying web application delivers forecasts, outperforming the NHC's statistical baseline for Atlantic storms. The open-source tool aims to aid both professionals and amateurs in tropical cyclone predictions, fostering better preparedness.

Global Synthetic Weather Radar Capability in Support of the U.S. Air Force (2019)

Haig Iskenderian, Mark S. Veillette, Christopher J. Mattioli, Patrick M. Lamey, Eric P. Hassey, Ashish Banerjee, Mark Worris, Kendrick Cancio, Shibi Rajagopalan, Hammad Usmani, John P. Dreher, Nessa Hock, John Radovan

The U.S. Air Force and MIT Lincoln Laboratory have collaborated to develop a synthetic weather radar capability, addressing global areas with limited weather data. This system utilizes a machine learning framework with inputs from global lightning, GALWEM numerical model, and weather satellite images to generate radar mosaics and forecasts up to 12 hours. It aims to enhance the Air Force's decision support systems. The presentation provides insights and outcomes from this partnership.

Multivariate LSTM Approach to Hurricane Intensity and Tracking Predictions (2021)

Akash B. Patel, Hammad Usmani, Jonathan C. Brant

In the backdrop of climate change and global warming intensifying hurricane conditions, there's an urgent need for real-time prediction of hurricanes and tropical storms. This research leverages deep learning, comparing a multivariate LSTM network with univariate LSTMs for hurricane prediction. The study utilizes data from IBTrACS version 4, hosted by NOAA. The research evaluates the effectiveness of Bidirectional LSTM networks and showcases the superiority of the multivariate model in predicting hurricane trajectories and intensities using MAPE. The outcome aids timely allocation of emergency resources and better preparation for adverse weather conditions.

Global Synthetic Weather Radar in AWS GovCloud for the US Air Force (2020)

Mark S. Veillette, H. Iskenderian, P. M. Lamey and co-authors

The US Air Force and MIT Lincoln Laboratory collaborate to produce a machine learning application creating global radar-like mosaics for pre-flight planning and execution. Drawing on various data sources, a convolutional neural network designs synthetic weather radar mosaics, integrated with GALWEM for up to 12-hour radar-forward forecasts. The system, in development for AWS GovCloud, capitalizes on extensive cloud compute and storage resources, feeding into Air Force decision aids like WxCC and AFW-WEBS viewer. The initiative represents a pioneering move to transit a mature ML system to AWS GovCloud. The project's facets include capability inception, AWS GovCloud development, training with feedback, and upholding ML "best practices" for sustained functionality post-transfer.

P.A.M. - Personal Assistant Machine

Loebner Prize 2017 Entry

A recurrent neural network large language model that is multilingual.

Hammad Usmani

Microsoft Certified: Azure AI Engineer Associate

This certifies expertise to Plan and manage an Azure AI solution, Implement decision support solutions, Implement computer vision solutions, Implement natural language processing solutions, Implement knowledge mining and document intelligence solutions, and Implement generative AI solutions.

Google Cloud Generative AI

Complete courses titled Introduction to Generative AI, Introduction to Large Language Models (LLM) and Introduction to Responsible AI. This certifies expertise of products including Vertex AI.

Harvard Business School Online: Entrepreneurship Essentials

Entrepreneurship Essentials is a 4-week, 30-hour online certificate program from Harvard Business School. Entrepreneurship Essentials introduces participants to the entrepreneurial journey from finding an idea to gaining traction in the marketplace to raising capital for a venture. Participants learn an overarching framework—People, Opportunity, Context, Deal—to evaluate opportunities, manage start-ups, and finance ventures.

Udacity Nanodegree Machine Learning Engineer

Machine learning represents a key evolution in the fields of computer science, data analysis, software engineering, and artificial intelligence.This program teaches how to become a machine learning engineer, and apply predictive models to massive data sets in fields like finance, healthcare, education, and more.

Harvard Business School Online: Credential of Readiness

Harvard Business School Online CORe (Credential of Readiness) is a 150-hour certificate program on the fundamentals of business from Harvard Business School. CORe is comprised of three courses—Business Analytics, Economics for Managers, and Financial Accounting—developed by leading Harvard Business School faculty and delivered in an active learning environment based on the HBS signature case-based learning model.

Big Data - Programming

The badge holder demonstrates the ability to use programming concepts provided by the various technologies in the Hadoop ecosystem including, but not limited to MapReduce and Pig.

Big Data Foundations

This badge holder has a basic understanding of Big Data concepts and their applications to gain insight for providing better service to customers. The learner understands that Big Data should be processed in a platform that can handle the variety, velocity, and the volume of data by using components that requires integration and data governance.

Big Data Hadoop Foundations

This badge holder has a basic understanding of Hadoop. The badge holder can describe what Big Data is and the need for Hadoop to be able to process that data in a timely manner. The individual can describe the Hadoop architecture and how to work with the Hadoop Distributed File System (HDFS) both from the command line and using the BigInsights Console that is supplied with IBM BigInsights.