Holberton School is a Silicon Valley-based computer training school that trains students to be full-stack software engineers and sees itself as an alternative to traditional 4-year college. Julien Barbier, co-founder and CEO of the newly formed school said that “traditional schools are great at teaching theory, but students don’t get much hands-on training. And, in the process, students spend almost fours years in college just to learn theory. When they go out to find jobs, companies that hire them first need to train them. These companies make a bet — sometimes it works, sometimes it doesn’t.”
Holberton School, by contrast, focuses on hands-on, project-based education. And it appears that the approach is working.
NASA selected Sravanthi Sinha, a first-year student at Holberton School, to participate in a summer workshop organized by the NASA Frontier Development Lab (FDL) and hosted at the SETI Institute.
From coding to saving earth
The six-week long research accelerator program, which ran from June 27 to August 5, brought together a team of postgraduate researchers in data analytics and planetary science and challenged them to think outside the box on the threat of asteroid impacts, according to the press release announcing Sinha’s selection into the program.
Bill Diamond, CEO of SETI, told CIO.com that Sinha was a highly qualified candidate in her own right, with her own background. “But her application was made more intriguing because of the Holberton School and its alternative approach.”
Diamond said that one of the things they liked about the school was its philosophy of applying real-world scenarios in team work and using all available resources to learn and get things done.
“This is a different approach than traditional computer science or software education and that has some intriguing element to it,” said Diamond. “I think the Holberton School, from an education perspective, is very much part of the Silicon Valley DNA in terms of saying we don’t have to stick to same way we have been doing things forever. We’ve got a different education model that I think is powerful and impactful and innovative.”
The FDL project goals were intense and the work had to be extremely collaborative. “There is collaboration between planetary scientists and data analytics so you got two different disciplines coming together and they have to quickly learn how each other thinks and operate and figure out how to collaborate to get something interesting done,” said Diamond. “It’s great to have a student from the Holberton School as part of the program. It’s an opportunity to see, from her own experience and from ours, how that educational platform has helped facilitate the work on the project.”
Sylvain Kalache, co-founder of Holberton School, told CIO.com that Sinha was part of the team that tackled the meteorite hunting problem, led by Dr. Peter Jenniskens, an expert on meteor showers.
Until now, various fireball networks (networks of cameras set up to observe meteors) have recorded approximately 800 trajectories of meteoroids significant enough to have dropped meteorites on the ground. In only 3 percent of cases have meteorites been recovered.
The solution the team proposed is to use drones to go in the impact area and survey the land to find the freshly fallen meteorites. The drone collects pictures that are then analyzed by an image recognition algorithm supported by a deep learning neural network.
Sinha built the web app that is in charge of collecting pictures and applying the deep learning algorithm on it to detect meteorites. As increasing the dataset is crucial for any deep learning algorithm, the web app will be publicly available to allow the community to contribute by uploading pictures. The web app will also allow humans to apply a meteoright or meteorwrong designation to pictures that have been flagged as potentially containing a meteorite.
Now the open source part…
The web app is using open source technologies such as PHP, Bootstrap, MySql and Apache and the team is working on making the web application code as well as the image recognition algorithm open source.