AI Data Program Manager

At Honeycomb, we're not just building technology , we’re reshaping the future of insurance.

In 2025, Honeycomb was ranked by Newsweek as one of “America’s Greatest Startup Workplaces,” and Calcalist named it as a “Top 50 Israel startup.”

How did we earn these honors?

Honeycomb is a rapidly growing global startup, generously backed by top-tier investors and powered by an exceptional team of thinkers, builders, and problem-solvers. Dual-headquartered in Chicago and Tel Aviv (R&D center), and with 5 offices across the U.S., we are reinventing the commercial real estate insurance industry, an industry long overdue for disruption. Just as importantly, we ensure every employee feels deeply connected to our mission and one another.

With over $55B in insured assets, Honeycomb operates across 18 major states, covering 60% of the U.S. population and increasing its coverage.

If you’re looking for a place where innovation is celebrated, culture actually means something, and smart people challenge you to be better every day - Honeycomb might be exactly what you’ve been looking for.

About The Role

As AI Data Program Manager, you’ll design and own the entire human-in-the-loop data pipeline — from defining annotation guidelines to ensuring label quality, managing workflows, and connecting feedback from production. You’ll work with underwriters, AI researchers, and product to ensure our training data reflects real-world complexity, yet remains consistent, scalable, and useful.

You’ll manage a growing operation of 20+ annotators, reviewers, and experts. You’ll define tasks that reduce subjectivity, improve reliability, and feed directly into the training of computer vision and vision-language models.

This is a hybrid role — part product, part ops, part cognition, and deeply involved in shaping how our AI sees the world.

What You’ll Do

• Translate expert underwriter knowledge into clear annotation tasks and ontologies
• Design and evolve annotation guidelines, workflows, and QA processes
• Manage and scale a team of expert and non-expert annotators
• Monitor data quality, coverage, and consistency across visual and multimodal tasks
• Collaborate with ML, data engineering, and product to define feedback loops and training datasets
• Own versioning and documentation for training/evaluation data
• Help close the loop between production outcomes (e.g., claims) and model training signals
  
  

• 4–7 years in data QA, human-in-the-loop annotation, research coordination, or ML data management (experience with computer vision is preferable)
• Experience with annotation tools (e.g., Encord, Labelbox, CVAT) and structured QA workflows
• Analytical mindset — detail-oriented, yet systems-focused
• Strong organizational, communication, and documentation skills
• Degree in a scientific, technical, or cognitive field such as Computer Science, Engineering, Cognitive Science, Experimental Psychology, Linguistics, HCI, or Information Systems
• Coursework or experience in statistics, data analysis, or machine learning fundamentals - an advantage
• Experience with multimodal environment (vision,text,tabular) - an advantage
  
  

• Lead the training data behind production-grade AI
• Influence cutting-edge models in insurance, vision, and GenAI
• Join a team where your thinking, structure, and decisions shape the core product
• High autonomy, impact, and opportunity to grow with the org