Resonant Exoplanets

Problem

Most exoplanets were identified manually, and AI-only detection can misclassify signals; for biosignatures, a false positive is the costliest mistake.

Solution

A full-stack pipeline that fetches Kepler, TESS and JWST data, detects transits with machine learning, validates candidates with physics models, and checks atmospheric chemistry with an exact-arithmetic solver to separate biosignatures (such as O₂ and CH₄ coexisting) from abiotic false positives, with plots and automatically generated scientific reports in a web interface.

Demo moment

Not described in the sources.

Why it stood out

• organizer statedWon the award for the project that best makes space data accessible or leverages it for a unique application.Source ↗
• organizer statedReplaces manual identification with automatic analysis and goes as far as possible biosignatures, using mission data such as JWST spectra.Source ↗
• team statedAims at the weak spot of AI-only detection — false positives — by adding physics validation and exact chemistry checks before calling anything a biosignature.Source ↗
• team statedShips the whole path rather than a model: data fetch, detection, validation, web interface and automated reports.Source ↗
• curator inferenceTook the challenge's task (find exoplanets) one step further to the question the public cares about (signs of life), raising both the stakes and the story.

Lessons for your next hackathon

• Push the challenge's task one step toward the question people actually care about.
• In AI challenges judged on validity, show how you catch the model's false positives.
• Name the mission data you used; validity judges look for it.