Cognify
Personal knowledge, made searchable.
A second brain that reads your documents. An asynchronous pipeline extracts, summarizes, classifies, and embeds files, with semantic search and cited answers across your vault.
Django · Next.js · Celery · Transformers
Inside Cognify
Document intelligence
Cognify is a full-stack personal knowledge management system — a second brain that reads your documents. Scanned PDFs, images, and plain-text files enter an asynchronous Celery worker pipeline.
OpenCV preprocessing and Tesseract OCR extract text. DistilBART summarizes it, a zero-shot MNLI classifier tags topics, and Sentence-BERT creates semantic embeddings for search and analysis.
A workspace for knowledge
The Next.js 16 dashboard brings together a knowledge vault, Recharts analytics, and browser-native PDF tools to split, merge, and reorder documents.
Document versioning preserves parent-child lineage. Sensitive mode, soft-delete recovery, collections, annotations, bookmarks, and notes keep documents and quotations organized.
Retrieval & answers
Cosine similarity search finds meaning across documents rather than relying only on keyword matches. In the global RAG chat, RoBERTa generates cited answers from the most relevant chunks in the vault.
Upload → OCR → Summarize → Classify → Embed → Completed
Access & operations
The Django REST backend enforces JWT authentication, user-scoped data isolation on every query, and audit logging for every action.
35 automated tests, Docker Compose orchestration, and Prometheus worker-latency metrics support the implementation. MIT licensed.
Built with Python, Django, Next.js, Celery, Transformers, Sentence-BERT, Docker, Prometheus
