The headline case. Spearheaded an LLM-powered insurance/finance knowledge database that became a user-facing GenAI assistant answering natural-language queries across desktop and mobile. Established QA metrics, adoption monitoring and iteration cycles; directed cross-functional engineering, data science, UX and product teams.
- 97%query accuracy
- 65%faster resolution
- +233%adoption / 6 months
A turn toward reliability — making AI features observable and trustworthy in production. Shipped user-facing ML pipeline monitoring dashboards; defined requirements for scalable ML pipeline products on containerized platforms; ran agile ceremonies.
- +15%development velocity
- ContainerizedML pipelines
Brought in to turn ML into product that scales. Partnered with business stakeholders to design and deliver ML-powered CRM and analytics platforms; translated OKRs into roadmaps; mentored engineers and data scientists.
- 100→40,000daily active users
- −35%latency
- +25%CRM effectiveness
The origin: a founder building a last-mile logistics SaaS with predictive analytics from zero. Owned product end to end — MVP scope, roadmap and market-fit via direct client engagement; architected microservices in GoLang/AWS with user-facing and delivery-person mobile apps.
- 0→1product shipped
- GoLang/AWSmicroservices