Full Stack Challenge: SavorPath – Map Your Culinary Experience
SavorPath is a personal food memory map for Indian foodies - log meals geographically and discover who you are through what you ate. Built solo with Amplify and AppSync, this article reveals why real-time sync matters for spatial memory and how one builder shipped a full-stack product in weeks.
The Problem
You snap a photo of the best biryani in Hyderabad. Six months later, you want it again. But:
- Instagram buried it under 1,000 other photos
- Google Maps forgot it
- MyFitnessPal reduced it to calories Every food app misses the same thing: your food isn't data. It's geography.
The Gap Nobody's Filling
Food memory is spatial. You think:
- "When I was in Bangkok"
- "That place in Mumbai"
- "Home on Sunday nights" NOT: "March 15th, 485 kcal"
I built SavorPath because a photo of incredible pad thai at midnight in Bangkok should be retrievable by place, not date. When the app shows you that photo months later, saying, "Remember this? Time to revisit?" - you actually book that trip.
That's the product.
The Three Things That Make It Work
1. The Map as Memory Palace
You log a meal with:
- Photo
- GPS location
- Mood tag (😋 Obsessed, 😌 Comfort, 🤯 Mind Blown, 😊 Casual) It becomes a pin on your flavor map. No timeline. No algorithm. Just geography.
Result: Zoom into your city and see every meal you've eaten there.
2. Flavor Fingerprint – Your Taste Identity
After 10 meals, a radar chart emerges showing:
- Spice Tolerance (0–100): How adventurous are you?
- Comfort Dependency (0–100): How often do you retreat to familiar food?
- Adventure Index (0–100): New restaurants per trip?
- Local Loyalty (0–100): How often do you revisit the same spots? Result: "I'm 72% comfort-food dependent at home, 90% adventure when traveling." That's insight. That's why you open the app weekly.
3. "Eat It Again?" Reminders
A notification: "Remember this? Time to revisit?"
One tap. You're back at the location. Save it. Plan a trip around it. Memory becomes action.
Why I Built This Solo
Indian Food Culture is Hyperlocal.
A Bangalore biryani enthusiast doesn't just "eat" - they remember:
- Which restaurant
- Which day
- Which owner
- They build relationships with places
They Want Ownership
Not Instagram visibility. Not Yelp ratings. A private archive that's theirs.
MVP is Narrow
The MVP requires a map, logging, and flavor fingerprint
AWS Amplify meant no server management, just product code.
Building Geographic UI: Four Real Problems
1. Reverse Geocoding Latency
Problem: GPS returns coordinates (13.0858° N, 80.2142° E). Users want "Chennai, Tamil Nadu." AWS Location Service is async.
Solution: Optimistic updates
- Show coordinates immediately
- Replace with place name when API responds
- Users expect this now (like Google Maps)
2. Mapbox Clustering at Scale
Problem: 500 meal pins on a map freeze the app.
Solution:
- Cluster aggressively at zoom < 8
- Expand to individual pins at Zoom 9+
- Mobile profiling exposed bottlenecks DevTools missed
3. Real-Time Sync
Problem: User A in Mumbai logs a meal. User B in Bangalore needs to see it instantly. No refresh. No "reload data?"
Solution: AppSync Subscriptions
- Push updates over WebSockets
- Each meal is a separate DynamoDB item
- Lambda validates → photo to S3 → meal to DB → AppSync broadcasts
- Real-time, no CRDTs needed
4. DynamoDB Geographic Queries
Problem: Can't range query latitude AND longitude simultaneously.
Solution: Denormalization
- Store
gridCellwith each meal ("lat_13_lng_80" for Chennai)
- Query by grid cell, not bounding box
- Trade-off: storage cost up, latency down
The Architecture
Browser (React 18 + Mapbox)
↓
Amplify Hosting → Cognito (Auth)
↓
AppSync (GraphQL + WebSockets)
├── DynamoDB (Meals, Profiles)
├── S3 + CloudFront (Photos)
└── Lambda (Validation, geocoding)Why This Stack?
Service | Why |
|---|---|
Amplify | No DevOps required |
AppSync | Real-time subscriptions built-in |
DynamoDB | Fast queries, no schema migrations |
S3 | Direct uploads, cheaper bandwidth |
Lambda | Scales to zero |
CloudFront | Global photo delivery (<100ms) |
Cognito | Auth without boilerplate |
Location Service | Reverse geocoding API |
What Shipping Taught Me
1. Build the Risky Core First
- Map sync working by day 2
- Everything else was polish
- If sync broke, the product failed
2. Managed Services Compound
- Cognito handled auth
- AppSync handled subscriptions
- Location Service handled geocoding
- Lambda handled jobs Each reclaimed hour. As a solo builder, that's shipping speed.
3. Geographic UI is Harder Than Expected
- Zoom levels
- Clustering algorithms
- Bounding box calculations
- Reverse geocoding latency
- Mobile rendering performance Weeks of iterative work.
4. Test on Real Devices Early
- Localhost Chrome DevTools lie.
- Old Android phones struggle at scale (500+ markers at zoom 7)
- Mobile profiling matters
Resources
Live App
- URL: savorpath.app
Source Code
Tech Stack
Frontend:
- React 18
- Mapbox GL
- Tailwind CSS Backend:
- Node.js Lambda
- TypeScript AWS Services:
- Amplify (Hosting, CI/CD)
- AppSync (GraphQL + WebSockets)
- DynamoDB (Meals, Profiles)
- S3 + CloudFront (Photos + global CDN)
- Cognito (Auth)
- Lambda (validation, geocoding, batch jobs)
- Location Service (Reverse geocoding)
- SNS (Notifications)
- CloudWatch Events (Daily reminder triggers)
Key Takeaways
✅ Geographic memory works differently than chronological logs
✅ Real-time sync requires WebSocket thinking, not polling.
✅ Managed services compound into shipping speed.
✅ Geographic UI complexity is underestimated.
✅ Solo builders benefit from a narrow MVP scope
Every bite tells a story. SavorPath makes sure you never forget it.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article