🛠 Tech Stack

💼 About This Role

You'll design and maintain evaluation benchmarks for vision models that parse unstructured enterprise documents like PDFs and spreadsheets. Your core impact will be surfacing model weaknesses and driving improvements. You'll collaborate closely with ML and GTM teams to shape how model quality is defined at a high-growth AI startup.

🎯 What You'll Do

• Design and maintain evaluation benchmarks to reveal model failure modes
• Develop metrics and heuristics to automatically identify new failure modes
• Partner with ML engineers to turn evaluation insights into model improvements
• Build lightweight internal and user-facing tools for inspecting model behavior

📋 Requirements

• Strong Python skills for building clean, reliable technical solutions
• Experience with data infrastructure such as AWS S3 and OLAP systems
• Ability to work hands-on with unstructured data (PDFs, spreadsheets)

✨ Nice to Have

• Experience at an early-stage or high-growth startup
• Product thinking and ability to build simple user-facing interfaces
• Background in AI/ML or document understanding

🎁 Benefits & Perks

• 🏖️ Unlimited PTO
• 🍱 Free daily lunch at the office
• 🚖 Reimbursed transportation costs
• 🏥 Generous health insurance (medical, dental, vision)
• 💪 $150/mo health and wellness budget