Datasaur

Datasaur, based in Livermore and Sunnyvale, CA, and also operating remotely, is a company in the B2B sector, specifically within the sub-industry of Engineering, Product, and Design. As a participant in the Y-Combinator W20 batch, Datasaur has developed a specialized platform for managing data labeling workforces, which is essential for natural language processing (NLP) tasks. The platform is designed to enhance the efficiency of human labelers through built-in intelligence that helps prevent costly errors and supports a wide array of NLP labeling tasks including named entity recognition, text classification, coreference resolution, and more. Datasaur’s recent innovations include a feature for building and training custom models for ChatGPT and other large language models (LLMs) in collaboration with LLM Labs. The company also offers robust API support for seamless integration with data pipelines on platforms like AWS and Azure, and supports a variety of text formats such as .csv, .txt, .pdf, and .ppt. Datasaur’s platform allows for multi-pass labeling and automatically calculates inter-annotator agreement, ensuring high-quality data annotation. The platform can be installed on clients’ own cloud servers with AES-256 encryption and is HIPAA and SOC2 compliant, making it a secure choice for Fortune 100 companies and those requiring air-gapped systems.