Chris Swart has six years of experience delivering Natural Language Processing services across the email, complaint, pharma, and sales industries. That breadth has given him a practical view of what it takes to move NLP from research demos to systems that create business value, especially in domains where language is messy and labels are expensive.

Chris is the cofounder of Comtura, a company on a mission to help sales teams use their customers’ voices to sell more, born from sales call transcription and CRM integration. There he leads a machine learning team of three, balancing hands-on modeling work with the everyday realities of shipping a product.

A recurring theme in his work is cost effective dataset creation. Chris focuses on distant supervision and semi-supervised approaches that get the most out of every labeled example, along with annotation workflows, model-assisted labeling, and quality metrics such as inter-annotator agreement. He shares this experience to help teams build high-quality NLP datasets without breaking their budgets.