Why We Built This
Two real frustrations, both ignored by existing tools:
- Space learning was scattered. Students were jumping between NASA, Wikipedia, and random videos just to understand one mission — wasting hours, still missing the full picture. We wanted one unified, AI-guided platform instead.
- Wellbeing tools only log, never respond. Every fatigue or mood tracker on the market just quietly stores your data — no one actually talks back to the tired, stressed person on the other end. We wanted an AI companion that notices and responds, in the user's own language (9 Indian languages), instead of handing them a dashboard of numbers nobody reads.
Both problems came from the same root: technology treating people as data points instead of human beings. OrbitX and MAITRI were our answer — built for SIH 2025, under team name Mediuzer, clearing the internal round.
The Story Behind It
It started at 2 AM, mid-study session — five tabs open just to understand one space mission, and somewhere in that same tired scroll, a thought about every "wellbeing" app that just silently records your stress and never says a word back. Two separate annoyances, one night, merging into a single question:
What if learning and caring both had software that actually paid attention?
That question became the pitch. OrbitX would fix the fragmentation — one place for planets, missions, telescopes, space tech, AI-guided instead of scattered. MAITRI would fix the silence — a companion that watches through your webcam, reads your emotion, and actually responds with empathy, music, and a voice in your own language.
Behind the Scenes
The unglamorous part — the part that doesn't make it into the pitch deck:
- Calibration bugs at 1 AM — the behavior detector kept flagging "Hyperactive" every time someone just leaned forward to grab water. Fixed by making it learn each person's natural baseline first, over the first 8 frames, before judging anything.
- The alert system originally spoke in pure engineering jargon — "CRITICAL: prolonged ocular closure detected" — when a tired, anxious person just needed to hear "hey, maybe rest your eyes for a bit." Had to rewrite the entire alert voice from scratch.
- WebSockets that loved dropping mid-demo — right when the webcam feed needed to be live in front of judges.
- The Echo Guard — a small, strange bug where MAITRI kept "hearing" its own voice and transcribing itself as user input. Built a dedicated guard just to stop the AI from talking to itself.
- Most of the real build happened in stolen hours — between college, hackathon prep, and very little sleep. Fix one module, break another, repeat — the actual rhythm of building anything real.
Cleared the SIH 2025 internal round on one conviction: build something that doesn't just measure people — it notices them.
🔗 GitHub for verification: https://github.com/ivengexnce/Space-Explorer-AI-Wellbeing