backend frontend devops ai Live
SentimentLens
Full-stack sentiment analytics — Next.js dashboard, FastAPI ML backend, and Kubernetes DevOps
Overview
Overview
SentimentLens ingests high-volume social streams and classifies brand sentiment in near real time. A fine-tuned DistilBERT model powers multi-class predictions while FastAPI and Redis keep the pipeline responsive under load.
Architecture
Posts enter through async workers, pass through preprocessing and inference, and stream results to dashboards via SSE. PostgreSQL stores aggregated metrics for historical reporting.
Key Highlights
- 94% accuracy on multi-class sentiment classification
- Processes 1M+ posts per day with async task queue
- Fine-tuned DistilBERT on proprietary brand dataset
- Real-time streaming via SSE for live dashboard updates
Tech Stack
PythonPyTorchHuggingFaceFastAPIRedisPostgreSQLDockerReactNext.jsKubernetes
Timeline
- Started
- 2024-07
- Status
- Ongoing