Creating AI-powered Knowledge Retrieval Applications using RAG
Budget: $250 – $750 USD
Secure AI Assistant App for Regulated Knowledge Fields (Education, Legal, Admin)
We are looking for a professional full-stack developer to build a secure mobile app using Flutter, with a custom backend (preferably using Strapi, Laravel, or Node-based CMS).
The app is designed for users in regulated and knowledge-based fields (e.g. education, legal, administrative) who seek precise answers through a private AI assistant.
All answers must be generated strictly from curated internal documents uploaded by the admin, using Retrieval-Augmented Generation (RAG) — not from internet or external sources.
- Core Features
1. AI Chat Assistant (RAG-Based)
• Uses RAG (Retrieval-Augmented Generation) architecture.
• Integrates with LLM (local model or secure API).
• Answers come only from uploaded content, with strict privacy — no internet access, no data sent to OpenAI or Gemini.
• Supports multi-turn chat, Arabic language, and dialect flexibility.
• Filters responses based on user role/classification.
• Stores chat history (local or cloud-based).
• Users can report incorrect answers, saved for admin review.
• Each answer must include clear source references (document name, section, page).
• System must validate answers post-generation to ensure they are within permissions and based on internal documents only.
2. User Management
• Registration via email or phone.
• Custom registration fields (e.g. job title, level...).
• Role-based access: User – Consultant – Admin.
• Users only access content allowed for their role or subscription.
3. Subscription & Access System
• Annual plans with auto-renewal alerts.
• Payment gateways: Stripe, Tap, PayPal.
• Auto-cancel on failed renewal; admin-controlled reactivation.
• Users may access premium documents by subscription or one-time purchase.
4. Document & Knowledge Base Management
• Upload & manage: PDF / DOCX / TXT files.
• Admin panel to categorize content by book/chapter/section.
• Auto-tagging on upload, with ability to manually edit tags.
• Real-time document updates via admin dashboard.
• File access controlled by role and payment status.
5. Digital Document Store & Book Viewer
• Users can buy/download protected files.
• Integrated book-style viewer with chapter-based navigation.
• No direct PDF viewing — content is structured dynamically.
• Smart linking between topics and sections.
6. Search & Filtering
• Full-text and keyword-based search.
• Filters by topic, document type, user role.
• Internal embedding + vector DB (e.g. Qdrant, Weaviate).
7. Notifications & Feedback
• Push notifications via Firebase or OneSignal.
• Admin can send custom notifications by role or event.
• Admin dashboard to review flagged answers and adjust content.
- Technology Stack (Suggested)
• Frontend: Flutter + Riverpod / BLoC / Provider
• Backend: Strapi / Laravel / Node.js (Headless CMS)
• AI Layer: Local LLM or secure OpenAI API (sandbox mode)
• Database: Firestore / SQLite + Vector DB (Qdrant / Weaviate)
• Notifications: Firebase / OneSignal
• Architecture: Modular, scalable, clean code, documented
- Project Security & Data Policy
• All content and user data must be stored on a private server or VPS.
• No data may be sent to any external LLM or third-party storage.
• The AI assistant must only generate responses from admin-uploaded content.
• System must allow data logging, admin audit, and content updates.
- Additional Enhancements
• Each AI response should include Source Preview (document name, section, page number).
• Post-processing validation must ensure responses match user permissions and come from internal sources only.
• Consider Express.js + Prisma as a lightweight backend option for MVP; upgrade to NestJS + Prisma later if scaling is needed.
- Deliverables
• Complete Flutter mobile app (Android + iOS)
• Admin dashboard (web-based)
• Clean, modular, documented code (backend + frontend)
• Deployment & hosting instructions
• Demo video and test APK
• Figma design (if custom UI is included)
- Timeline & Payment
Please provide:
• Timeframe (including milestone breakdown)
• Cost estimate per stage (design, backend, frontend, AI integration, testing)
• Payment schedule linked to deliverables
Note to Freelancers:
Please submit a realistic and final budget based on the full scope of the project.
Also, kindly share a link or sample of a similar project you have completed.
We are looking for a professional full-stack developer to build a secure mobile app using Flutter, with a custom backend (preferably using Strapi, Laravel, or Node-based CMS).
The app is designed for users in regulated and knowledge-based fields (e.g. education, legal, administrative) who seek precise answers through a private AI assistant.
All answers must be generated strictly from curated internal documents uploaded by the admin, using Retrieval-Augmented Generation (RAG) — not from internet or external sources.
- Core Features
1. AI Chat Assistant (RAG-Based)
• Uses RAG (Retrieval-Augmented Generation) architecture.
• Integrates with LLM (local model or secure API).
• Answers come only from uploaded content, with strict privacy — no internet access, no data sent to OpenAI or Gemini.
• Supports multi-turn chat, Arabic language, and dialect flexibility.
• Filters responses based on user role/classification.
• Stores chat history (local or cloud-based).
• Users can report incorrect answers, saved for admin review.
• Each answer must include clear source references (document name, section, page).
• System must validate answers post-generation to ensure they are within permissions and based on internal documents only.
2. User Management
• Registration via email or phone.
• Custom registration fields (e.g. job title, level...).
• Role-based access: User – Consultant – Admin.
• Users only access content allowed for their role or subscription.
3. Subscription & Access System
• Annual plans with auto-renewal alerts.
• Payment gateways: Stripe, Tap, PayPal.
• Auto-cancel on failed renewal; admin-controlled reactivation.
• Users may access premium documents by subscription or one-time purchase.
4. Document & Knowledge Base Management
• Upload & manage: PDF / DOCX / TXT files.
• Admin panel to categorize content by book/chapter/section.
• Auto-tagging on upload, with ability to manually edit tags.
• Real-time document updates via admin dashboard.
• File access controlled by role and payment status.
5. Digital Document Store & Book Viewer
• Users can buy/download protected files.
• Integrated book-style viewer with chapter-based navigation.
• No direct PDF viewing — content is structured dynamically.
• Smart linking between topics and sections.
6. Search & Filtering
• Full-text and keyword-based search.
• Filters by topic, document type, user role.
• Internal embedding + vector DB (e.g. Qdrant, Weaviate).
7. Notifications & Feedback
• Push notifications via Firebase or OneSignal.
• Admin can send custom notifications by role or event.
• Admin dashboard to review flagged answers and adjust content.
- Technology Stack (Suggested)
• Frontend: Flutter + Riverpod / BLoC / Provider
• Backend: Strapi / Laravel / Node.js (Headless CMS)
• AI Layer: Local LLM or secure OpenAI API (sandbox mode)
• Database: Firestore / SQLite + Vector DB (Qdrant / Weaviate)
• Notifications: Firebase / OneSignal
• Architecture: Modular, scalable, clean code, documented
- Project Security & Data Policy
• All content and user data must be stored on a private server or VPS.
• No data may be sent to any external LLM or third-party storage.
• The AI assistant must only generate responses from admin-uploaded content.
• System must allow data logging, admin audit, and content updates.
- Additional Enhancements
• Each AI response should include Source Preview (document name, section, page number).
• Post-processing validation must ensure responses match user permissions and come from internal sources only.
• Consider Express.js + Prisma as a lightweight backend option for MVP; upgrade to NestJS + Prisma later if scaling is needed.
- Deliverables
• Complete Flutter mobile app (Android + iOS)
• Admin dashboard (web-based)
• Clean, modular, documented code (backend + frontend)
• Deployment & hosting instructions
• Demo video and test APK
• Figma design (if custom UI is included)
- Timeline & Payment
Please provide:
• Timeframe (including milestone breakdown)
• Cost estimate per stage (design, backend, frontend, AI integration, testing)
• Payment schedule linked to deliverables
Note to Freelancers:
Please submit a realistic and final budget based on the full scope of the project.
Also, kindly share a link or sample of a similar project you have completed.
Related categories:
Backend Development
Flutter
NLP
Vector Databases
Retrieval-Augmented Generation (RAG)