VRL RAINA ELUMALAI E

JourneyGraph AI — Evidence-Linked Digital Identity & Intelligent Memory System

🎯 PROJECT OVERVIEW JourneyGraph AI transforms fragmented academic and professional documents into a searchable, evidence-backed digital identity. Students can upload their certificates, resumes, internship letters, project reports, and transcripts — the system automatically understands, categorizes, and connects everything into a unified intelligent profile. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ ⚙️ SETUP STEPS (< 3 minutes) 1. Clone or download the project 2. Install dependencies: pip install streamlit pymupdf python-docx chromadb google-genai python-dotenv pydantic 3. (Optional) Add Gemini API key for AI features: Create a .env file in the project root: GEMINI_API_KEY=your_key_here Get a free key at: https://aistudio.google.com/app/apikey 4. Run the app: streamlit run app.py 5. Open browser at: http://localhost:8501 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 🔑 TEST CREDENTIALS No login required. The app is single-user. AI Mode: Requires a Gemini API key in .env Fallback Mode: Works 100% without any API key (keyword-based extraction) ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 🧪 HOW TO TEST 1. Go to "Upload & Analyse" in the sidebar 2. Upload any PDF, DOCX, TXT, PNG, or JPG file (Try: a certificate PDF, a resume, or a project report) 3. Click "Process All Documents" — watch real-time processing 4. Navigate to "Digital Identity" to see extracted skills + categories 5. Go to "Smart Search" — type "What skills do I have?" or "List my certifications" 6. Check "Journey Timeline" for chronological events 7. Visit "Relationship Graph" to see document-skill connections 8. See "Career Insights" for category distribution + growth tips ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 🏗️ KEY DESIGN DECISIONS 1. ZERO-CRASH FALLBACK ARCHITECTURE Every AI-powered function has a rule-based/keyword fallback. The app NEVER crashes because of a missing API key. 2. EVIDENCE-LINKED IDENTITY Every skill claim, organization mention, and achievement is linked back to its source document. No unverified claims. 3. LOCAL-FIRST, PRIVACY-SAFE All files stored locally in storage/uploads/ ChromaDB vector store is local (storage/chroma/) SQLite database is local (storage/journeygraph.db) Nothing is sent to any cloud without explicit API key. 4. DUAL-MODE INTELLIGENCE • With Gemini API Key: Full AI classification, extraction, OCR, semantic QA, structured metadata, embeddings • Without API Key: Keyword classifiers, regex extractors, BM25-style chunked search via ChromaDB's built-in embeddings 5. VECTOR SEARCH + RAG Documents are chunked, embedded, and stored in ChromaDB. Smart Search uses RAG (Retrieval-Augmented Generation) — retrieves relevant chunks, then generates grounded answers. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 📁 PROJECT STRUCTURE journeygraph-ai/ ├── app.py ← Main dashboard (home page) ├── pages/ │ ├── 1_Upload_and_Analyse.py ← Document ingestion pipeline │ ├── 2_Digital_Identity.py ← Evidence-backed profile │ ├── 3_Smart_Search.py ← RAG semantic search │ ├── 4_Journey_Timeline.py ← Chronological view │ ├── 5_Relationship_Graph.py ← Knowledge graph │ ├── 6_Document_Library.py ← File manager │ └── 7_Career_Insights.py ← AI insights ├── services/ │ ├── database.py ← SQLite CRUD │ ├── document_parser.py ← PDF/DOCX/TXT/Image text │ ├── classifier.py ← Category detection │ ├── ai_extractor.py ← Metadata extraction │ ├── embedding_service.py ← Gemini embeddings │ ├── vector_store.py ← ChromaDB wrapper │ ├── search_service.py ← RAG search + answer │ ├── relationship_engine.py← Graph relationships │ ├── timeline_service.py ← Chronological events │ └── identity_service.py ← Profile generation ├── models/ ← Pydantic data models ├── utils/ ← File, date, text helpers └── storage/ ← Local DB + uploads + vectors ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 🏆 WHAT MAKES IT UNIQUE • Every skill in your profile is backed by a source document citation • Upload images — Gemini Vision reads text from scanned certificates • Works 100% offline (fallback mode) — no internet or API needed • Real-time processing pipeline with progress visualization • Vector semantic search over YOUR private documents only • Career gap analysis — tells you what evidence you're missing