Level 03 · Innovator
Real problems, real impact.
Innovator isn’t more classes. It’s a studio: pick a real problem, scope it, build a working solution with the tech you already know, ship it to real users, learn what they think. Eight project briefs to choose from, or pitch your own.
How innovators work
Three frameworks that scaffold the process
Innovation isn’t inspiration — it’s a repeatable process. We use three established frameworks across every project: design thinking from Stanford and IDEO, and MIT’s AI-app methodology.
Stanford d.school K-12 Design Thinking
The originator of the five-stage process: Empathize → Define → Ideate → Prototype → Test. Free method cards (How Might We, Crazy 8s) adapted for K-12.
IDEO Design Thinking for Educators Toolkit
IDEO’s teacher-facing companion to the d.school method. A 94-page Designer’s Workbook covering Discovery → Interpretation → Ideation → Experimentation → Evolution. Strong on project management.
MIT App Inventor + MIT RAISE Day of AI
For projects that ship as an app or use AI. App Inventor gets you to a working Android app in a weekend; Day of AI layers in the ethics and societal-impact lens.
Choose your project
Eight ready-to-start project briefs
Each brief is sized for a team of 2–4 students over 6–12 weeks. Pick one, modify one, or bring your own — the format works for any project that solves a real problem for real people.
AI + Mobile
Grandparent Translator
- The problem
- Kids whose grandparents speak limited English struggle to have real conversations across generations.
- Who benefits
- Multi-generational immigrant families.
- Tech stack
- Python + Whisper (open-source STT) + free-tier LLM API + Streamlit web app or MIT App Inventor mobile app.
- Good-enough v1 (6 weeks)
- Phone-held app that listens, transcribes, translates one direction, reads back aloud.
- Stretch v2 (12 weeks)
- Hands-free two-way conversation mode with a pre-loaded “family glossary” (names, foods, inside jokes) so translations stay personal.
- Real-world precedent
- Google Live Translate on Pixel phones is the consumer analog. No specific student precedent located.
3D printing + Innovation
Senior Adaptive Grips Library
- The problem
- Arthritic seniors struggle with door knobs, keys, pill bottles, zippers. Commercial aids are expensive and not personalized.
- Who benefits
- Seniors at a partner senior center or assisted-living home.
- Tech stack
- Tinkercad or Fusion 360 + FDM 3D printer (PLA).
- Good-enough v1 (6 weeks)
- Three printed grip attachments fitted to one real senior after an in-person measurement visit.
- Stretch v2 (12 weeks)
- A Printables/Thingiverse collection of 8–10 designs with a sizing guide so any senior center can print them.
- Real-world precedent
- e-NABLE volunteer network for 3D-printed prosthetic hands (enablingthefuture.org). Same model, different body part.
IoT + Civic Data
School Air Quality Dashboard
- The problem
- Poor classroom ventilation increases illness and reduces cognitive performance. Admins lack data.
- Who benefits
- Students and teachers in the school.
- Tech stack
- Raspberry Pi or ESP32 + SDS011 PM2.5 sensor + SCD40 CO₂ sensor + Python + free-tier Grafana Cloud dashboard.
- Good-enough v1 (6 weeks)
- One sensor in one classroom, posting CO₂ and PM2.5 to a public dashboard once per minute.
- Stretch v2 (12 weeks)
- Five sensors across the building + automated email alerts above 1000 ppm CO₂ + written report to the principal.
- Real-world precedent
- “CO2 Monitor in Schools” Italian citizen-science project (co2.mysen.org); CMU CREATE Lab Speck program (cmucreatelab.org).
AI + Citizen Science
Backyard Bird AI Counter
- The problem
- Cornell’s eBird needs broader data. Local kids rarely contribute scientifically usable observations.
- Who benefits
- Citizen science (Cornell Lab of Ornithology) and the team’s neighborhood.
- Tech stack
- Raspberry Pi + USB mic + BirdNET-Pi (open source) + Python script that posts daily counts.
- Good-enough v1 (6 weeks)
- Working BirdNET-Pi in one student’s backyard logging detections for two weeks.
- Stretch v2 (12 weeks)
- Three nodes across different habitats (yard, park, schoolyard) with a comparison dashboard and an eBird submission.
- Real-world precedent
- BirdNET-Pi itself; Cornell Lab’s BirdNET model (birdnet.cornell.edu) is used by thousands of hobbyists.
AI + Civic Information
Community Events Summarizer
- The problem
- Local school, library, and city event info is scattered across five websites and a Facebook page. Busy parents miss things.
- Who benefits
- Families in the school district.
- Tech stack
- Python web scraper (BeautifulSoup) + free-tier LLM API + weekly email via free Mailchimp tier or a static site on GitHub Pages.
- Good-enough v1 (6 weeks)
- Manually triggered weekly digest summarizing events from 3 sources, sent to 10 test families.
- Stretch v2 (12 weeks)
- Auto-runs every Sunday night, covers 8+ sources, includes a “kid-friendly events only” filter, opt-in signup.
- Real-world precedent
- Code for America brigades have built similar civic info aggregators (brigade.codeforamerica.org). No specific youth precedent located.
Accessibility + AI
Color-Blind Helper for the Art Room
- The problem
- About 8% of boys are colorblind. Art class, science diagrams, and graphs assume full color vision.
- Who benefits
- Colorblind students at the school.
- Tech stack
- MIT App Inventor mobile app + phone camera + open-source color-naming library or a small TensorFlow Lite model.
- Good-enough v1 (6 weeks)
- Point-the-camera-at-a-thing app that speaks the dominant color name out loud.
- Stretch v2 (12 weeks)
- Adds a “simulate what a colorblind classmate sees” mode for teachers, plus a graph-recoloring tool that re-renders class slides with colorblind-safe palettes.
- Real-world precedent
- Color Blind Pal iOS app; Microsoft Seeing AI (microsoft.com/en-us/ai/seeing-ai).
AI + Education
Peer Tutoring AI for 6th-Grade Math
- The problem
- Younger students at the school struggle with fractions and pre-algebra. Teachers can’t give 1:1 time.
- Who benefits
- 6th graders at the same school.
- Tech stack
- Free-tier LLM API + Streamlit web app + custom system prompt loaded with the school’s actual curriculum + guardrails (hints only, never final answers).
- Good-enough v1 (6 weeks)
- Web app that walks a student through one chapter using the Socratic method, tested with 5 sixth graders.
- Stretch v2 (12 weeks)
- Full semester coverage, logs which concepts trip students up most, weekly report for the math teacher.
- Real-world precedent
- Khanmigo (khanmigo.ai) is the production version of exactly this idea.
AI + 3D printing + Civic
Senior Emergency Kit Advisor
- The problem
- Seniors living alone often lack a current emergency kit and don’t know what’s missing.
- Who benefits
- Seniors at a partner senior center.
- Tech stack
- MIT App Inventor app or printed checklist + a small LLM-powered web form that customizes the kit based on medications, mobility, and pets + 3D-printed labeled bin dividers.
- Good-enough v1 (6 weeks)
- Interview 5 seniors, generate personalized kit checklists, deliver printed binders.
- Stretch v2 (12 weeks)
- Self-serve web tool + a “kit check” reminder system that texts a family member every 6 months.
- Real-world precedent
- No specific precedent located. Red Cross and FEMA publish generic checklists, but personalization by youth teams appears original.
Pitch your own
Got a different problem in mind?
These eight briefs are starting points. If your team sees a problem worth solving that isn’t on this list — pitch it. We’ll help scope it.