Space Apps / NASA data projects
Questions worth exploring.
Ocean observations and lunar life support offer different ways to learn about connected systems. Open a project, inspect its evidence and see where its conclusions stop.
NASA Space Apps NYC 2025
FINDS: Sharks From Space
Edmund Gunn Jr. and Yasmine Dweir
Explore how ocean temperature and chlorophyll observations can support questions about marine habitat. The map displays exploratory environmental scores with source provenance.
Project history
The 2025 NYC hackathon prototype received the local Best Use of Gemini API award. Subsequent v2.x work adds the NASA observation pipeline, bounded AI scoring, tests, cloud caching and Android packaging. Those later capabilities are separate from the original prototype and award.
How it works
The documented v2.x pipeline samples NASA GIBS MUR sea-surface temperature and PACE/VIIRS chlorophyll visualizations, decodes official colormaps and applies an explicit environmental heuristic. Gemini ranks and explains existing geographic cells. A Cloudflare Worker and R2 cache serve the React/Deck.gl map. The repository documents a web PWA and Capacitor Android app with prior Pixel 6a evidence.
What you can investigate
Inspect the source date and environmental layers, then explain what a high score can and cannot tell you. Engineers can follow the pipeline, structured output guards, architecture diagrams and claim/evidence matrix.
Scientific and validation limits
Environmental suitability scores are not verified shark sightings, real-time tracking, encounter-risk forecasts or marine-safety warnings. The project does not implement a validated species-distribution model. Its shark-tag design remains conceptual, with no deployed tag or live tag telemetry. Prior automated Pixel evidence does not establish human usability sign-off or support for untested platforms.
NASA Space Apps Chicago 2026
7GC Mission Control: Lunar Outpost Resilience Trainer
Solo Chicago entry: Edmund Gunn Jr., gunnchOS. AI assistance: OpenAI Codex.
Keep two astronauts supplied during a fictional 24-hour lunar exercise. Repair an array fault, shelter the crew and choose how much power and water to devote to science and greenhouse care.
Project history
Built for Build a Junior Astronaut Mission Trainer. The V1.1.0 competition checkpoint was submitted on November 15, 2026 at 5:43 AM America/Chicago. This native branded host preserves the tested application. No 2026 award is claimed. The independent entry remains separate from the broader gunnchOS production ecosystem.
How it works
One deterministic engine drives play, forecasts, exact replay and a constrained search of 567 phase policies. In the reference scenario, 199 preserve authored reserves and 18 preserve them across five fixed sensitivity cases. The balanced plan gives 9 science hours and 12 greenhouse-care hours. The selected all-case plan retains a minimum 1.035 kWh battery in the combined adverse case. A simulated radio outage demonstrates a bounded queue, deferred science delivery and explicit overflow loss. Forty-one model/planning tests and browser checks accompany the source.
What you can investigate
Junior, Explorer and Engineer views share one state. Explain which activity your plan sacrificed, compare alternatives and inspect the units and trace. The mission and planner also run from one standalone offline HTML file.
Scientific and validation limits
NASA NTRS Tables 2, 8 and 11 anchor dry-food, water, oxygen and electrolysis inputs. Hardware, greenhouse, radio, reserves and exposure points are teaching assumptions. Greenhouse care produces no one-day harvest. The model omits clinical dose, full habitat chemistry, thermal dynamics and physical RF propagation. It is not NASA-certified, and classroom learning outcomes and assistive-technology usability still need evaluation.
Independent participant projects. Use of NASA publications or observations does not imply NASA endorsement.