Internal AI Tools

Building AI Agents That Save Hours Across the Product Lifecycle

Building the internal AI tooling that speeds up Yahoo's product and engineering workflows.

• Built an AI-powered experiment review agent using Claude Code — flagging anomalous data and generating first-draft experiment reviews, cutting review cycle time by roughly 20%.
• Developed agentic dashboards using Claude and SQL — automatically generating plain-language narratives of product health metrics, saving 5+ hours of manual reporting weekly.
• Shipped AI-powered API workflows that automate query categorization — uncovering three key model-tagging errors in production data and enabling engineering teams to ship targeted model improvements.
• Delivered an AI-powered Success Metrics Builder using Claude — independently prototyped, validated with real users, and deployed into the product workflow, automating the synthesis of stakeholder success criteria.