_edited_edited_edited.png)
The Question
There are millions of hours of deep-sea video footage in public ocean research archives, recorded by underwater robots, baited camera traps, autonomous vehicles, going back decades. Until recently, processing that footage meant a graduate student watching frame by frame, classifying what they saw, missing whatever they weren't looking for. AI image processing has changed what's possible. The questions ocean scientists can now ask at scale: what lives in the deep ocean, in what abundance, and how is that changing? That is what Dr. Sandoval's students will work on.
What Students Build
Over five weeks, students learn how ocean image classification models actually work, run existing models on real deep-sea video data from public ocean research archives, evaluate where those models succeed and where they fail, and explore whether they can do better, either by combining outputs from multiple models or by fine-tuning on specific ocean species.
The work is real research methodology. Students don't just use AI tools, they evaluate them. They learn to spot where a model is confident but wrong, where it's uncertain but
right, and what that distinction means for ocean science. Dr. Sandoval brings the domain expertise to validate outputs and guide which questions matter. The Academy team supports the technical implementation.
Students leave with a working evaluation pipeline, an analysis of how current ocean image AI performs on real data, and the kind of artifact ocean scientists actually use in their research. They will also have seen, on screen, more deep-sea life than most people see in a lifetime.
The Mentor
Dr. Jessica Sandoval is an ocean engineer who designs deep-sea remotely operated vehicles and tagged sperm whales as a Harvard postdoc on Project CETI, the effort to build a Rosetta Stone for whale communication. She has turned her attention to making deep-sea technology accessible beyond the research vessel and the PhD. This summer her students will use AI to do what graduate students have spent entire careers doing one frame at a time: process underwater video footage to ask what lives in the deep, in what abundance, and how that is changing.
Who This Is For
Python proficiency is required, comfort with libraries, file handling, and basic data manipulation. Some exposure to image processing or machine learning is helpful but not required. The Academy team will scaffold the technical implementation. The right student here is genuinely curious about the ocean, watches deep-sea footage and wants to know what they're seeing, and is comfortable with the slow grind of working with real data that doesn't always cooperate. Students who need a polished outcome at every step will struggle. Ocean science moves at its own pace.
Logistics
Five weeks. July 6 to August 7, 2026. Mondays, Wednesdays, and Fridays, 1:00 PM to 2:15 PM ET. Friday sessions are Demo Days. Cohorts of 3 to 4 students per mentor. $4,500. Apply by May 25, 2026 at 11:59 PM.
Beyond the live sessions, students work on their own, and they are not alone when they do. The lab is supported by a 24/7 Slack channel and a team of scholars and practitioners at the Academy. Students also work alongside SeqHub's AI co-teacher, which helps them think through problems on off days without doing the work for them. Plan for 10 to 12 hours per week, with 4.5 hours in live sessions and the rest on independent work.