Josh Veitch-Michaelis

I'm a computer vision engineer interested in instrumentation and applying machine learning for social good.

prof_pic.jpg

I’m currently in Antarctica working as a winterover for the 10-m South Pole Telescope, a radio instrument observing echoes of the very early universe.

Previously, I was a postdoc jointly between the DS3Lab at ETH Zurich (Systems Group), supervised by Prof. Ce Zhang, and Restor Eco. My research was funded by a Google.org impact grant project to explore how best to apply ML to datasets that Restor has collected (imagery, audio data, LIDAR, etc.). This culminated in the release of OAM-TCD, a large dataset of very high resolution aerial images and labels for tree crown delineation. This dataset is one of the largest and most diverse of its kind, specifically constructed to facilitate low-cost, scaleable, global tree mapping. I’m also on the Insights team for BioDivX, a multi-institution group of researchers competing in the Rainforest X Prize (we made it to the finals and are waiting to see how we did!)

The last time I wintered at Pole was for the University of Wisconsin, Madison for the IceCube Neutrino Observatory, where I lived for just over a year in 2020-2021.

For a while, I’ve worked at the intersection of academia and industry. Some of the more interesting projects include: putting ML models in space to detect flooding, building intelligent drone payloads for endangered species conservation, maintaining the world’s largest neutrino telescope, designing 3D imaging systems for extreme environments and trying to automate spectral calibration for astronomers everywhere.

I obtained my PhD from the Mullard Space Science Laboratory (University College London) in 2016, which won the Remote Sensing and Photogrammatery Society’s annual award for best PhD thesis, as well as a SIM scholarship from the Worshipful Company of Scientific Instrument Makers for excellence in instrumentation research.

Me, taking pictures of the aurora australis in front of the South Pole Telescope late in winter 2024 - photo by my co-winterover Kevin Zagorski