I love working on compelling software engineering and machine learning opportunities that matter. When i'm not working towards my December 2018 graduation, I'm working on my company - Viden Technologies - and my ongoing Tuberculosis research project - LIRA.
What started with tinkering with my calculator has since progressed to my lifelong goal of making the world a better place. Please don't hesitate to reach out to me with interesting opportunities in software engineering!
Collaborated with other technical engineers and product managers in an agile development environment to create a microservice architecture capable of automatically detecting and resolving errors in financial data
Generated projected cost savings of 75M per year for U.S. clients and 500K per year for IBM using this framework
Created an assistive image analysis system to recognize and analyze Tuberculosis lesions using machine learning and neural networks
Saving these solutions in a repository of machine learning and computer vision models for other labs to leverage in similar biological and medical research problems, such as oncology
Working with 9+ national labs and diagnostic facilities to develop custom assistive software for biological and medical research
Lowered the level of expertise needed to classify lesions, enabled higher granularity of lesion statistics, and increased the speed at which drug efficacy tests can be accomplished by over 4X
Made use of Bayesian Statistical models to create a tool capable of automatically finding and tuning optimal hyperparameters for expensive-to-train machine learning models, such as deep convolutional neural networks
Wrote a detailed tutorial and blog post describing the method, and applied the optimizer to a variety of machine learning models
Developed a denoising algorithm for multiclass classification problems, with inspiration from conventional smoothing and denoising algorithms, for use in development of the Tuberculosis Lesion Recognition project
Applied this algorithm to drastically improve results in the Tuberculosis Lesion Recognition project, by removing extraneous misclassifications throughout our output predictions