Semantas
About
Semantas is an advanced AI-powered text annotation and relevant data extraction system designed to transform how enterprises handle natural language processing and information extraction tasks. The platform leverages sophisticated NLP algorithms and machine learning models to automatically analyze text documents, identify key entities, extract relevant information, and annotate content with unprecedented accuracy and efficiency. Built with enterprise requirements in mind, Semantas features robust SSO authentication for seamless integration with corporate identity management systems, combined with reCAPTCHA security measures to prevent unauthorized access. The platform's flexible architecture supports various use cases across industries, from legal document review and healthcare records analysis to financial reporting and customer feedback processing. By automating the tedious and error-prone process of manual data annotation, Semantas significantly reduces operational costs, improves data quality, and enables organizations to derive actionable insights from their textual data assets at scale.
Tech Stack
Key Features
- AI-powered automatic text annotation with entity recognition and information extraction capabilities
- Custom NLP model training interface for domain-specific text classification and data extraction tasks
- Batch processing pipeline for analysing large document volumes with progress tracking and result export
- SSO authentication integration for seamless enterprise identity management and access control
- Interactive annotation review interface with confidence scoring and manual correction tools
My Role
Developed the AI-powered NLP annotation platform using FastAPI for the Python backend and Vue.js for the frontend. Built the text processing pipeline with Python NLP libraries, implemented the SSO authentication layer, and created the interactive annotation review interface with confidence scoring and batch export capabilities.
Challenges & Solutions
The main challenge was building a performant NLP pipeline that could process large document batches in real-time while providing interactive feedback to users through the Vue.js frontend. This required implementing async task processing with FastAPI's background tasks, streaming annotation results via WebSocket for live progress updates, and optimising the Python NLP models for production throughput without sacrificing extraction accuracy.
