Kyle Dieruf

Leading Marketplace Search and AI: the company’s core search experience, dealer products, and Carson, its first AI search agent.

• Led delivery of Carson, Cars.com’s AI search agent, recognized with a 2026 AI Excellence Award: 6–8M unique users per month, 25–30% adoption among web searchers, and roughly 20–25% of all web leads, with Carson users converting at meaningfully higher rates from search results to vehicle pages to dealer leads. Delivered by a 7-person team of 4 engineers and 3 data scientists.
• Led the shift to agentic software delivery, starting with my Search and AI teams and scaling to peer teams across the Marketplace organization, 8 teams in total. Teams ground agents in organizational context (an approach I call Intent Engineering), turn workflows into reusable agent skills, and enforce quality through evals, CI, and policy-as-code. Agents now author nearly every pull request, cycle time has dropped from weeks to hours or days, and every product manager and designer on these teams ships code.
• Shipped the Cars.com ChatGPT App to production, establishing the company’s first public-facing MCP server and a live presence inside ChatGPT. Delivered without dedicated engineering resources, from stakeholder alignment through AgentCore infrastructure.
• Drove the formation of a cross-functional AI Engineering team, merging application engineers with Data Science to eliminate organizational silos, and coached Data Science leadership to adopt a rapid, iterative delivery model alongside engineering.
• Defined the engineering strategy for Carson, including the Vector DB vs. API-Orchestrated GenAI analysis that set its hybrid architecture by showing why vector-only retrieval breaks the ecommerce contract.
• Led the end-to-end rebuild of the Search Results Page on a server-driven UI (SDUI) architecture across web, iOS, and Android, delivering 14x faster page loads (LCP from 24s to 1.67s at p95).
• Authored an SEO Execution Plan that turned weeks of bespoke engineering per search pattern into hours of self-service configuration, letting the SEO team operate independently of engineering cycles.
• Earned 100% positive feedback from my direct reports in the company engagement survey, with every project commitment met and the team operating autonomously through a 10-week manager absence.