SentiGenix
Real-time sentiment analysis with VADER NLP classification and AI-guided text rewriting
The Problem
Teams often publish text with unintended tone. The gap was a tool that can classify sentiment quickly and propose rewrites while preserving meaning.
Architecture
System:React -> Django REST API -> VADER NLP + DeepSeek API
React -> Django REST -> VADER (classification) + DeepSeek (rewriting)
Key decisions:
- VADER for low-latency classification
because VADER provides near-instant sentiment feedback suitable for interactive typing workflows. - Two-stage pipeline
because Separating classify and rewrite keeps fast feedback independent from LLM generation latency.
Impact
- Shipped and deployed at sentigenix.abhinavchaurasia.in
- Sub-5ms sentiment classification via VADER
- DeepSeek rewrite suggestions preserving key facts
What I’d Do Differently
I’d add a domain-specific fine-tuned classifier and stream rewrite output for more responsive UX.