$wap

Challenge Statement

In Arkansas, thousands of people have talent, but limited access to opportunity. Small businesses make up 99.3% of all enterprises (≈ 258,000), yet the state still ranks 40th out of 51 in overall economic activity.

Rising costs mean many small business owners skip professional services — not because they lack vision, but because every dollar goes toward staying open. At the same time, college students and freelancers across Little Rock possess valuable skills they can’t fully use or afford to market.

The problem isn’t a lack of ability. it’s a lack of access.

$wap addresses this by asking:

How might we help communities unlock value without relying on money?
How might we turn “what you can do” into “what you need”?
How might AI connect people so skills never go unused?

Project Description

$wap is an AI-powered skill-exchange platform that lets people trade skills instead of cash. It creates a new, inclusive economy where your abilities — not your wallet — open doors.

How It Works

Create a Profile — List the skills you can offer and what you need.
AI Matchmaking — Machine learning pairs you with users whose skills and availability fit yours.
Swap Fairly — Exchange services, time, or expertise — no money required.
Earn & Redeem Credits — Every successful swap earns credibility and “swap credits.”
Build Trust — Ratings and feedback verify quality and accountability.

Example

A college student in Little Rock who codes websites can trade with a hairstylist who needs a site.
She builds the website; he braids her hair.
No cash — just community, fair value, and mutual growth.

Why It Matters to Little Rock and Arkansas

Small Business Owners often skip marketing or tech services because every dollar counts.
College Students need real-world experience but can’t afford mentorship or tools.
Freelancers & Creators have skills that sit idle without capital or visibility.

By turning skills into currency, $wap activates Arkansas’s underused talent and fosters a local economy built on collaboration and trust — not cash.

Unique Value Proposition

Traditional freelance platforms require money, subscriptions, and transaction fees.
$wap reimagines value exchange as reciprocity, not currency.

No middlemen — only fair exchange.
No barriers — anyone can start with what they know.
No limits — skills build community wealth.

For small businesses, it means affordable growth.
For freelancers, it means stronger portfolios and networks.
For college students, it means real-world experience without debt.

Tech Stack

Website / Frontend: Flutter (web) — responsive UI, single codebase targeting web/desktop/mobile.

Frontend

Built with Flutter & Dart

Tech Stack:

Framework: Flutter 3.9+ with Material Design 3
Language: Dart
State Management: StatefulWidgets with setState
Authentication: Firebase Auth with Google Sign-In
Database: Cloud Firestore for user profiles and data persistence
Storage: Firebase Storage for media assets
HTTP Client: http package for backend API communication
3D Rendering: model_viewer_plus for interactive 3D models (.glb assets)
Responsive Design: Adaptive layouts for mobile, tablet, and desktop

Key Features:

🔐 Authentication: Email/password and Google OAuth sign-in
🔍 Semantic Search: Real-time search with timeout handling and retry logic
🎨 Modern UI: Dark theme with purple accent, gradient buttons, and smooth animations
📱 Responsive: Works seamlessly across web, iOS, and Android
🎯 3D Models: Interactive ModelViewer components for engaging visuals
🚀 Performance: Optimized asset bundling and lazy loading

Architecture Highlights:

Service-based architecture (AuthService, SearchService, ProfileService)
Reusable widget components (AppSidebar, custom form fields)
Centralized theme management with Material 3 ColorScheme
Error handling with user-friendly snackbar notifications
Web-safe CORS handling (omits Authorization header on web to avoid preflight)

Deployment:

The frontend is deployed on Netlify for fast, reliable hosting with global CDN distribution.

Deployment Process:

Build: Run flutter build web --release to generate optimized production assets
Deploy: Use netlify deploy --prod --dir build/web to push to production
CDN: Assets are automatically distributed across Netlify's global CDN
SSL: HTTPS enabled by default with automatic certificate management

Live URL: https://swap-besmart.netlify.app

Why Netlify?

⚡ Instant Deploys: Changes go live in seconds
🌍 Global CDN: Fast loading times worldwide
🔒 Secure: Automatic HTTPS and DDoS protection
🔄 CI/CD Ready: Can integrate with GitHub for automatic deploys
📊 Analytics: Built-in deployment and performance monitoring
💰 Free Tier: Generous free tier perfect for MVP deployment

Configuration for Flutter Web:

Assets are bundled in build/web/assets/assets/ (double nesting)
3D models (.glb files) referenced with assets/assets/ prefix for web compatibility
Firebase OAuth configured with Netlify domain in authorized domains list
Responsive breakpoints: mobile (<780px), tablet (780-1200px), desktop (>1200px)

Backend

Built with FastAPI and machine learning for smart skill matching.

Python 3.11 + FastAPI - REST API
Firebase Firestore - User profile storage
Qdrant Cloud - Vector database for semantic search
Redis - Caching layer (16x faster repeat queries)
sentence-transformers (BERT) - Text embedding model
Docker - Containerization
Fly.io - Production hosting

Backend Features:

Semantic search using BERT (finds skills by meaning, not just keywords)
Reciprocal matching algorithm (finds mutual exchanges)
Skill recommendation engine (suggests complementary skills)
Fast vector similarity search (~80ms, ~5ms with cache)
Redis caching layer (16x speedup on repeat queries)
384-dimensional embeddings for each skill
Harmonic mean scoring for balanced matches

Production API: https://swap-backend.fly.dev

Full Backend Documentation

Authentication & Database

Firebase Authentication - Secure user management
Firebase Firestore - Profile data storage with automatic indexing
Qdrant Cloud - Vector storage for 10,000+ profiles (free tier)
Redis - Cache layer (local dev only, not on production yet)

Architecture

flowchart TB
  UI[Flutter Web UI] --> API[FastAPI Server]
  API --> PM[Profile Management]
  API --> SS[Semantic Search]
  API --> RM[Reciprocal Matching]
  API --> ML[ML Embeddings]
  ML --> QD[Qdrant]
  PM --> FB[Firebase Firestore]
  SS --> QD
  SS --> FB
  RM --> QD
  RM --> FB
  API --> RD[Redis Cache]
  
  classDef dark fill:#0F0F11,stroke:#8B5CF6,stroke-width:2px,color:#FFFFFF;
  class UI,API,PM,SS,RM,ML,FB,QD,RD dark;

Data Flow

flowchart LR
  UI["User Input"] --> API["FastAPI Validation"]
  API --> FB["Firebase (store profile)"]
  API --> ML["ML Model (encode skills)"]
  ML --> QD["Qdrant (store vectors)"]

  classDef dark fill:#0F0F11,stroke:#8B5CF6,stroke-width:2px,color:#FFFFFF;
  class UI,API,FB,ML,QD dark;

Semantic Search:

flowchart LR
  Query["User Query (e.g. 'guitar lessons')"] --> Encode["ML Model (encode -> 384-dim)"]
  Encode --> QD["Qdrant (vector search)"]
  QD --> FB["Firebase (fetch full profiles)"]
  FB --> Results["Ranked Results (by cosine similarity)"]

  classDef dark fill:#0F0F11,stroke:#8B5CF6,stroke-width:2px,color:#FFFFFF;
  class Query,Encode,QD,FB,Results dark;

Reciprocal Matching:

flowchart LR
  A[My Skills] --> EncodeA[Encode A]
  B[My Needs] --> EncodeB[Encode B]
  EncodeA --> SearchA[Search Their Offers]
  EncodeB --> SearchB[Search My Offers]
  SearchA --> ScoreA[Score A]
  SearchB --> ScoreB[Score B]
  ScoreA & ScoreB --> Harmonic[Harmonic Mean]
  Harmonic --> Top[Top Mutual Matches]
  
  classDef dark fill:#0F0F11,stroke:#8B5CF6,stroke-width:2px,color:#FFFFFF;
  class A,B,EncodeA,EncodeB,SearchA,SearchB,ScoreA,ScoreB,Harmonic,Top dark;

Performance

Operation	Cached	Uncached	Components Used
Semantic Search	~5ms	~80ms	FastAPI → ML → Qdrant → Firebase
Reciprocal Match	~8ms	~120ms	FastAPI → ML → Qdrant (2x) → Firebase
Profile Create	-	~150ms	FastAPI → Firebase → ML → Qdrant
Profile Read	-	~20ms	FastAPI → Firebase
Health Check	-	~1ms	FastAPI only

Tested on: Fly.io (1GB RAM, 1 CPU), 50+ profiles

Note: Redis caching speeds up repeat queries by ~16x. Currently only running in local dev (docker-compose), not deployed to production yet.

API Documentation

Backend REST APIs

Production Base URL: https://swap-backend.fly.dev
Local Base URL: http://localhost:8000
Interactive Docs: https://swap-backend.fly.dev/docs (Swagger UI)

1. Health Check API

Endpoint: GET /healthz

Verify backend services are operational.

Request:

curl https://swap-backend.fly.dev/healthz

Response: 200 OK

{
  "status": "healthy",
  "firebase": "connected",
  "qdrant": "connected"
}

2. Profile Management API

Create/Update Profile

Endpoint: POST /profiles/upsert

Create a new user profile or update existing one with skill embeddings.

Request Body:

{
  "uid": "sarah_j_2024",
  "email": "[email protected]",
  "display_name": "Sarah Johnson",
  "skills_to_offer": "I can teach web development with React and JavaScript. I also help people build their first website from scratch.",
  "services_needed": "Looking for someone who can teach me product photography and basic photo editing skills.",
  "bio": "Software engineer who loves building things",
  "city": "Little Rock"
}

Response: 200 OK

{
  "uid": "sarah_j_2024",
  "email": "[email protected]",
  "display_name": "Sarah Johnson",
  "skills_to_offer": "I can teach web development with React and JavaScript. I also help people build their first website from scratch.",
  "services_needed": "Looking for someone who can teach me product photography and basic photo editing skills.",
  "bio": "Software engineer who loves building things",
  "city": "Little Rock",
  "created_at": "2025-11-08T10:30:00.000Z",
  "updated_at": "2025-11-08T10:30:00.000Z"
}

What happens:

Profile stored in Firebase Firestore
Skills text converted to vectors using BERT
Vectors indexed in Qdrant for search

Try it:

curl -X POST https://swap-backend.fly.dev/profiles/upsert \
  -H "Content-Type: application/json" \
  -d '{
    "uid": "marcus_williams",
    "email": "[email protected]",
    "display_name": "Marcus Williams",
    "skills_to_offer": "I can help you learn how to play bass guitar and understand music theory basics",
    "services_needed": "Want to learn mobile app development, especially Flutter"
  }'

Get Profile

Endpoint: GET /profiles/{uid}

Retrieve someone's profile.

Response: 200 OK

{
  "uid": "tyler_designs",
  "email": "[email protected]",
  "display_name": "Tyler Mitchell",
  "skills_to_offer": "I can teach graphic design using Figma and Adobe Illustrator. I also do logo design and brand identity work.",
  "services_needed": "Need help learning Spanish for an upcoming trip to Barcelona",
  "bio": "Creative designer in Fayetteville",
  "city": "Fayetteville",
  "created_at": "2025-11-07T14:22:00.000Z",
  "updated_at": "2025-11-08T09:15:00.000Z"
}

Try it:

curl https://swap-backend.fly.dev/profiles/tyler_designs

3. Semantic Search API

Endpoint: POST /search

Search using natural language. Works by meaning, not just keywords

Search Modes:

offers: Find people who can teach you something
needs: Find people who want to learn what you know
both: Search everything

Request:

{
  "query": "someone who can help me understand how to make beats and produce music",
  "mode": "offers",
  "limit": 5
}

Response: 200 OK

[
  {
    "uid": "dj_carlos",
    "display_name": "Carlos Rodriguez",
    "email": "[email protected]",
    "skills_to_offer": "I teach music production using FL Studio and Ableton. I can show you how to make trap, hip-hop, and R&B instrumentals.",
    "services_needed": "Want to get better at video editing, especially transitions and color grading",
    "bio": "Producer and DJ, been making beats for 8 years",
    "city": "Conway",
    "score": 0.87
  },
  {
    "uid": "melody_davis",
    "display_name": "Melody Davis",
    "email": "[email protected]",
    "skills_to_offer": "Can teach you music theory, sound design, and how to mix your own tracks professionally",
    "services_needed": "Looking for someone to teach me content writing and copywriting",
    "city": "Jonesboro",
    "score": 0.82
  }
]

Score ranges from 0 to 1 (higher = better match).

More examples:

Looking for a language tutor:

curl -X POST https://swap-backend.fly.dev/search \
  -H "Content-Type: application/json" \
  -d '{
    "query": "someone who speaks French fluently and can teach me conversational French",
    "mode": "offers",
    "limit": 5
  }'

Want to teach your coding skills:

curl -X POST https://swap-backend.fly.dev/search \
  -H "Content-Type: application/json" \
  -d '{
    "query": "I know JavaScript and React, looking for people who want to learn web development",
    "mode": "needs",
    "limit": 5
  }'

General search:

curl -X POST https://swap-backend.fly.dev/search \
  -H "Content-Type: application/json" \
  -d '{
    "query": "cooking and baking skills",
    "mode": "both",
    "limit": 8
  }'

4. Skill Recommendation API

Endpoint: POST /search/recommend-skills

Get personalized skill recommendations based on what you already know. Analyzes profiles with similar skills and suggests complementary skills that are commonly learned together.

Algorithm:

Encode your current skills using BERT
Search for profiles with similar skill sets
Extract skills from those profiles (both what they offer and need)
Rank by frequency and relevance score
Return top recommendations

Request:

{
  "current_skills": "Python programming and web development",
  "limit": 5
}

Response: 200 OK

[
  {
    "skill": "SQL databases and data modeling",
    "score": 0.845,
    "reason": "Common among 12 similar profiles"
  },
  {
    "skill": "Docker containerization and deployment",
    "score": 0.732,
    "reason": "Common among 8 similar profiles"
  },
  {
    "skill": "Git version control and collaboration",
    "score": 0.698,
    "reason": "Common among 10 similar profiles"
  }
]

Try it:

curl -X POST https://swap-backend.fly.dev/search/recommend-skills \
  -H "Content-Type: application/json" \
  -d '{
    "current_skills": "JavaScript and React development",
    "limit": 5
  }'

Features:

Cached for 2 hours for fast repeat queries
Smart scoring balances frequency and semantic relevance
Filters out vague or too-short skills
Useful for skill gap analysis and learning path planning

5. Reciprocal Matching API

Endpoint: POST /match/reciprocal

Find mutual skill exchange partners where you teach them and they teach you.

Algorithm:

Search their offers vs your needs → Score A
Search your offers vs their needs → Score B
Calculate harmonic mean: 2 × (A × B) / (A + B)
Rank by combined score

Why Harmonic Mean? Penalizes one-sided matches. Both scores need to be high.

Good match: (0.9, 0.9) → 0.90
One-sided: (0.9, 0.3) → 0.45

Request:

{
  "my_offer_text": "I can teach web development, specifically building sites with React and handling databases",
  "my_need_text": "Want to learn acoustic guitar, maybe some basic music theory too",
  "limit": 10
}

Response: 200 OK

[
  {
    "uid": "brandon_strings",
    "display_name": "Brandon Hayes",
    "email": "[email protected]",
    "skills_to_offer": "I teach acoustic guitar, electric guitar, and music theory for beginners and intermediate players",
    "services_needed": "Really want to learn web development so I can build my own website for my music lessons",
    "bio": "Guitarist in Hot Springs, been playing for 12 years",
    "city": "Hot Springs",
    "score": 0.91
  },
  {
    "uid": "jasmine_codes",
    "display_name": "Jasmine Parker",
    "email": "[email protected]",
    "skills_to_offer": "Can teach you piano basics and how to read sheet music",
    "services_needed": "Need help understanding how to code and build web applications",
    "bio": "Piano teacher trying to transition to tech",
    "city": "Bentonville",
    "score": 0.78
  }
]

Try it:

curl -X POST https://swap-backend.fly.dev/match/reciprocal \
  -H "Content-Type: application/json" \
  -d '{
    "my_offer_text": "I can help with graphic design and teach you Adobe Photoshop and Illustrator",
    "my_need_text": "Looking to learn how to bake bread and pastries from scratch",
    "limit": 10
  }'

API Response Codes

Code	Meaning	Description
200	OK	Request successful
400	Bad Request	Invalid input (validation error)
404	Not Found	Profile not found
500	Internal Server Error	Server-side error

External Services & APIs Used

1. Firebase Services

Firebase Authentication - User identity management
Firebase Firestore - NoSQL document database
- Collection: profiles
- Auto-indexing on uid, email
- Real-time sync capabilities

2. Qdrant Cloud API

Service: Vector similarity search database
Plan: Free tier (1GB storage, ~10,000 profiles)
Features Used:
- Named vectors (offer_vec, need_vec)
- HNSW indexing for fast search
- Cosine similarity matching
- REST API over HTTPS

3. Hugging Face Transformers

Model: sentence-transformers/all-MiniLM-L6-v2
Purpose: Convert text to semantic embeddings
Output: 384-dimensional vectors
Features:
- Pre-trained on semantic similarity tasks
- Normalized embeddings
- Fast inference (~10-20ms per encoding)

4. Redis

Service: In-memory data store for caching
Deployment: Local development only (via docker-compose)
Features:
- Sub-5ms response times
- 1 hour TTL on cached results
- Automatic cache invalidation on profile updates
- Graceful degradation if unavailable

5. Fly.io Platform

Service: Cloud application hosting
Region: US East (Virginia)
Features:
- Auto-scaling
- Global CDN
- HTTPS by default
- Health checks

Complete API Documentation with Postman Examples

User Stories

As a User, I want to sign up with email/password or Google OAuth so I can quickly create an account and access the platform.
As a Creator, I want to post my skills with a clear title, description, duration, mode (remote/in-person), and tags so others can discover what I can offer.
As a Seeker, I want to search for skills using semantic search so I can find people who offer what I need to learn, even with natural language queries.
As a User, I want to browse curated skill cards on the Discover page so I can explore available offerings without searching.
As a User, I want to request a skill swap from someone so we can connect and exchange services.
As a User, I want to manage my profile with my skills to offer and services needed so others can find me through search.
As a User, I want to see my requests in a dedicated dashboard so I can track ongoing and completed exchanges.

Walkthrough: Using $wap

Quick Start Guide:

1. Sign Up / Sign In

Visit: https://swap-besmart.netlify.app
Choose Sign Up to create a new account
Options:
- Email/password authentication
- Google Sign-In (one-click OAuth)
Fill in your profile details (name, skills to offer, services needed)

2. For Skill Providers (Creators)

Click "Post a Skill" from the sidebar
Fill in the skill details:
- Title: Clear, descriptive name (e.g., "React Hooks Masterclass")
- Category: Select from Design, Coding, Writing, Language, Music, etc.
- Description: What you'll teach and what learners will gain
- Duration: Estimated hours for the exchange
- Mode: Remote, In-person, or Both
- Tags: Add relevant keywords for discoverability
Click "Publish" to make your skill live
Your skill appears on the Discover page with verified badge (if applicable)

3. For Skill Seekers (Buyers)

Browse: Explore the Discover page to see curated skill cards
Search: Use the semantic search bar to find specific skills
- Example queries: "learn guitar", "web development help", "design logo"
- Backend uses AI embeddings to match your query with relevant skills
Filter: Use category chips (All Skills, Design, Coding, etc.)
Preview: Click on any skill card to see:
- Full description
- Provider rating
- Duration and mode
- Tags
Request: Click the "Request" button to initiate a skill swap
Fill in your request details and submit

4. Managing Requests

Click "Requests" in the sidebar
View all incoming and outgoing requests
Track status: Pending, Accepted, Completed
Communicate with swap partners
Mark completed exchanges

5. Search & Discovery

Top Search Bar: Quick search from any page
Discover Page Search: Detailed search with filters
Clear Results: Click the "X" or "Close" button to return to default view
Search results show:
- User profiles with matching skills
- Relevance score
- Skills offered and needed
- Bio and location

Demo/Test Credentials

For testing without creating a new account:

Email: [email protected]

Password: Test123!

Link to Demo Presentation

Please Click Link To View $wap Presentation Slide Deck

Team Checklist

Project Checklist

Presentation complete and linked (placeholder)
Code merged to main branch

School Name

Philander Smith University

Team Name

Panthers

Contributors

Immanuella Emem Umoren
Kenna Agbugba
Otito Udedibor
Olaoluwa James-Owolabi
Emmanuella Turkson

Name		Name	Last commit message	Last commit date
Latest commit History 53 Commits
.github/workflows		.github/workflows
swap_frontend		swap_frontend
wap-backend		wap-backend
.DS_Store		.DS_Store
.gitignore		.gitignore
README.md		README.md

BE-Hackathon-2025/Panthers

Folders and files

Latest commit

History

Repository files navigation

$wap

Challenge Statement

Project Description

How It Works

Example

Why It Matters to Little Rock and Arkansas

Unique Value Proposition

Tech Stack

Frontend

Backend

Authentication & Database

Architecture

Data Flow

Performance

API Documentation

Backend REST APIs

1. Health Check API

2. Profile Management API

Create/Update Profile

Get Profile

3. Semantic Search API

4. Skill Recommendation API

5. Reciprocal Matching API

API Response Codes

External Services & APIs Used

1. Firebase Services

2. Qdrant Cloud API

3. Hugging Face Transformers

4. Redis

5. Fly.io Platform

User Stories

Walkthrough: Using $wap

1. Sign Up / Sign In

2. For Skill Providers (Creators)

3. For Skill Seekers (Buyers)

4. Managing Requests

5. Search & Discovery

Demo/Test Credentials

Link to Demo Presentation

Team Checklist

Project Checklist

School Name

Team Name

Contributors

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors 3

Uh oh!

Languages

Packages