Tyler Zobel: AI Data Analyst

Atmospheric Soaring Technologies (Atmosoar) is a leading climate change startup utilizing big data to improve predictions. Specifically, when working with aerial technologies like drones, climate change is making it more difficult to predict when is a good time to operate. This summer, I led the evaluation and integration of AI into regional weather prediction, selecting and implementing a deep-learning architecture combining transformers and graph neural networks that reduced compute time from up to two hours to less than six minutes.

Notably, I was able to rapidly expand my capacity in data analytics: I preprocessed 30,000-plus cloud images from multiple datasets and developed PyTorch models to classify cloud types and assess precipitation potential, achieving greater than 85% validation accuracy. However, while evaluating the AI models for production, I realized that my interests in technology isn’t simply in producing a new data analytics model: it lies in the communication. Working with business experts and programmers alike, I found myself watching the transaction of information between two worlds, and I hope to find a place where I can continue that work professionally.

Internship funded by: CXD Funded Experiences and Skills