Each system begins with a stubborn vertical workflow and is engineered through production — combining industry context, conversation, reasoning, data, and action.
Selected work · 01Automotive
Vehicle diagnostics company
AI transformation of vehicle diagnostics and repair guidance
Technicians relied on printed manuals and tribal knowledge to diagnose complex vehicle faults. Diagnostic errors were costly, and repair times varied widely across service locations.
Modernized the diagnostics platform with ML models trained on vehicle telemetry, fault codes, and repair history. Delivered an AI copilot that guided technicians with ranked root-cause recommendations and step-by-step repair instructions.
- 35% faster average diagnostic time
- 28% reduction in repeat repairs
- Consistent repair quality across franchise locations
Selected work · 02Real Estate
Real estate management firm
AI operations for resident services and on-site workforce
Property managers were overwhelmed by resident complaints, maintenance requests, and manual insurance checks for contractors. Response delays hurt retention and compliance risk was rising.
Deployed an AI phone agent to answer and triage resident complaints, an intelligent work-management system to route and schedule maintenance, and automated insurance verification for on-site workers before each shift.
- 60% of resident calls resolved without human handoff
- Average work-order completion time cut by 40%
- 100% contractor insurance verification at shift start
Selected work · 03Automotive
Automotive marketing company
AI-powered automation of vehicle marketing campaigns
Campaigns for hundreds of dealerships were built manually. Creative, targeting, and media spend were inconsistent, and vehicles stayed on lots longer when listings missed the right buyers.
Built an AI marketing engine that auto-generated listings, personalized creative by audience, and optimized media spend across channels. The platform ingested inventory, pricing, and local buyer signals to launch campaigns automatically.
- 3x faster campaign launch time
- 22% improvement in lead-to-sale conversion
- 40% lower cost per qualified lead
Selected work · 04Technology / Startups
AI customer-support startup
AI-native support MVP from concept to launch
A seed-stage startup had a strong idea for an AI support agent but needed a working product in market fast. They had no AI team, no data infrastructure, and limited runway to hire.
Designed and built the MVP in eight weeks: a RAG-based agent integrated with Slack and email, an evaluation harness, a feedback loop for answer improvement, and a cost-tuned inference architecture.
- MVP live with paying design partners in 8 weeks
- 78% of routine questions resolved without human escalation
- Architecture ready to grow with the business