███████╗██╗ ██╗ ██╗██╗ ██╗
██╔════╝██║ ██║ ██║╚██╗██╔╝
█████╗ ██║ ██║ ██║ ╚███╔╝
██╔══╝ ██║ ██║ ██║ ██╔██╗
██║ ███████╗╚██████╔╝██╔╝ ██╗
╚═╝ ╚══════╝ ╚═════╝ ╚═╝ ╚═╝
Smart Resource Allocation for Social Impact
Turning Unstructured NGO Data into Intelligent Action
| Field | Details |
|---|---|
| Team Name | FLUX |
| Team Leader | Utkarsh Rastogi |
| Problem Statement | Smart Resource Allocation |
NGOs across India collect vast amounts of field data - surveys, reports, and assessments - yet most of it is never processed or acted upon. The result is misallocated resources, unmet urgent needs, and volunteers operating without clear direction.
FLUX is built to solve exactly this.
FLUX is an end-to-end intelligent resource allocation platform that transforms raw, unstructured NGO data into prioritized tasks and automatically routes them to the most suitable volunteer.
RAW DATA --> OCR EXTRACTION --> LLM ANALYSIS --> TASK GENERATION --> SMART ALLOCATION
| Challenge | FLUX's Solution |
|---|---|
| Unstructured NGO data piling up | OCR + LLM pipeline that processes any format |
| No prioritization of urgent needs | Urgency Detection Engine with real-time scoring |
| Wrong volunteer for the wrong task | Skill-based intelligent matching algorithm |
| System does not improve over time | Feedback loop that retrains and optimizes continuously |
- End-to-End Automation - From upload to execution, zero manual steps
- Hybrid AI Engine - Rules-based urgency detection combined with LLM reasoning
- Task Generation Engine - Our core innovation; converts raw insights into structured tasks
- Self-Improving Architecture - Gets smarter with every completed task via feedback
| Feature | Description |
|---|---|
| 📤 Multi-Format Upload | Accepts Images, PDFs, and plain text |
| 🔍 OCR Extraction | Converts scanned documents to structured data |
| 🧠 LLM Analysis | Gemini-powered contextual understanding |
| 🚨 Urgency Detection | Flags and prioritizes critical needs automatically |
| 🛠️ Task Generation | Auto-creates actionable tasks from raw insights |
| 👥 Smart Volunteer Matching | Assigns volunteers by skills, proximity, and availability |
| 📊 Admin Dashboard | Real-time oversight, reporting, and insight visualization |
| 🗺️ Volunteer Map View | Location-aware task discovery for volunteers |
| 🔁 Feedback System | Continuous learning from task outcomes |
| ☁️ Cloud Storage | Secure, scalable storage on AWS S3 |
Step 1 -> Admin uploads NGO data (Images / PDFs / Text)
Step 2 -> Data stored securely in AWS S3
Step 3 -> OCR Engine extracts raw text content
Step 4 -> Gemini LLM analyzes content and detects urgency
Step 5 -> Task Generation Engine creates structured tasks
Step 6 -> Matching Algorithm assigns the best-fit volunteer
Step 7 -> Volunteer executes task via mobile app
Step 8 -> Feedback captured and system improves
+---------------------------+
| Flutter App |
| (Admin + Volunteer UI) |
+-------------+-------------+
|
+-------------v-------------+
| AWS EC2 + FastAPI |
| Backend |
+--------+----------+--------+
| |
+-------------v--+ +----v--------------+
| AWS S3 | | OCR Engine |
| (Storage) | | (Text Extraction) |
+----------------+ +--------+-----------+
|
+----------------v-----------+
| Gemini LLM API |
| (Analysis + Task Gen) |
+----------------+-----------+
|
+---------------------+-v-----------+
| PostgreSQL + Firebase Firestore |
| (Data Layer) |
+------------------------------------+
| Layer | Technology |
|---|---|
| Frontend | Flutter & Dart |
| Backend | Python, FastAPI |
| Cloud | AWS EC2, AWS S3 |
| Database | PostgreSQL, Firebase Firestore |
| AI / LLM | Google Gemini API |
| Auth & Realtime | Firebase |
| Maps | OpenStreetMap |
| State Management | Riverpod |
- Upload raw NGO data in any format
- Dashboard with AI-generated insights and reports
- Task creation and volunteer oversight
- Map-based task discovery
- Task acceptance and rejection with one tap
- Real-time progress tracking and feedback submission
| Phase | Milestones |
|---|---|
| Phase 1 - Current | Core pipeline, smart matching, Admin + Volunteer app, feedback loop |
| Phase 2 | Resource Treasury Meter, Advanced Analytics, Gamification & Leaderboards |
| Phase 3 | Predictive Need Analysis, Multi-NGO Collaboration, National Scale Deployment |
Vision: Build a fully intelligent, self-optimizing NGO ecosystem that operates with minimal human intervention and maximum social impact.
| Resource | Link |
|---|---|
| 📂 GitHub Repository | github.com/urastogi2048/Flux |
| 🎥 Demo Video | Watch on YouTube |
| 🚀 MVP Build | Download MVP |
| 🧪 Prototype | View Prototype |
Made with ❤️ for Social Impact - Team FLUX