Dmitriy Karpman

CTO and Co-founder Hive

Last Reviewed: September 26, 2026

Executive Summary

Dmitriy Karpman is the co-founder and Chief Technology Officer of Hive, a San Francisco company that sells cloud-hosted, pre-trained AI models to enterprises and platforms. Hive’s models classify images, video, audio and text for content moderation, detect AI-generated content and deepfakes, and measure brand and sponsorship exposure; the company says they serve billions of API requests a month. As CTO he represents the technical side of a business built on the claim that models trained on large, carefully labelled datasets beat what customers could build themselves, a case he has argued in public at the computer vision research community’s main conference.

Career Highlights

• Co-founded Hive in 2017 with Kevin Guo, who serves as chief executive.
• Named by Hive as its CTO in June 2023, and listed as CTO and Co-founder of Hive among invited speakers at the first Workshop on Multimodal Content Moderation at CVPR 2023.
• At that workshop, presented Hive’s work on how training data quantity and label noise affect model performance, the use of synthetic data where real data is scarce, and how to find and correct bias in trained models.
• Included by Built In in its April 2025 list of the 45 most disruptive leaders in the artificial intelligence industry.

Professional Journey

Hive was founded in 2017. TechCrunch, reporting its $85 million in combined Series C and D funding in April 2021, described it as a pivot from a Q&A platform called Kiwi, which grew out of a project at Stanford. Built In credits Karpman as co-founder, alongside Kevin Guo, and reported in 2025 that the company had raised more than $120 million and grown to over 170 employees across offices in San Francisco and Delhi.

The company started with image classification and built up a portfolio of models trained on data labelled by a large distributed workforce. By 2021 TechCrunch counted around 100 customers, among them Reddit, Chatroulette, NBCUniversal, Interpublic Group, Walmart and Visa. Hive’s main markets are platform integrity and content moderation, advertising and sponsorship measurement, and, more recently, detection of AI-generated images, text, audio and deepfakes.

The clearest public record of Karpman’s technical work is his talk at the Workshop on Multimodal Content Moderation, held with the IEEE Conference on Computer Vision and Pattern Recognition in Vancouver on 18 June 2023. Hive’s write-up of the talk, which identifies him as its CTO, covers the research questions behind its models: how much the quantity and cleanliness of training data matter, whether synthetic data can stand in for real data, and how to detect and address bias once a model is trained. The experiments it describes, such as training an NSFW image classifier on between 10 and 100,000 images with up to half the labels deliberately corrupted, reflect the company’s argument that data quality is its main technical advantage.

Education

No education record was confirmed from a primary source. Secondary research profiles report undergraduate study at the University of Missouri-Columbia and a doctorate from Stanford, but neither could be checked against a first-party or institutional record, so neither is recorded here.