Ina
Edstrom

I trained an AI to pole vault — and it discovered a technique nobody had tried before.

That's what I do: combine technical rigor with genuine curiosity to find answers no one expected. As a machine learning researcher and national champion athlete, I approach every problem with the same mindset — push past the conventional, be precise, and keep going until something clicks.

My thesis used Proximal Policy Optimization to teach a virtual agent to pole vault from scratch — no prior sport knowledge, no predefined technique. It found its own way, ultimately clearing 6.76 meters with a side-swinging method never seen before. That result was 50 cm above the current world record.

I hold a Master's and Bachelor's in Computer Science from the University of South Alabama, a published paper in the 2022 IEEE Symposium Series on Computational Intelligence, and years of hands-on research in machine learning for both government and academic projects.

Outside of research, I'm a Swedish ACTFL-certified speaker, a lifelong athlete, and someone who believes the best solutions come from asking questions nobody else thought to ask.