AI / LATENCY / CONCURRENCY
Jiehong Li
AI PRODUCT ENGINEER / TECH LEAD
SELECTED WORK
One platform problem, one product built from nothing, and one change to how a team works.
Led the modernization of a real-time AI roleplay platform for enterprise-scale use. By separating the voice pipeline into independent services and moving session state to Redis, we scaled the system reliably from 500 to 10,000 RPS on Kubernetes.
Helped take exaBase Empower from concept to production, building the React experience, Django APIs, and the LLM pipeline behind its HR analytics. The product launched in September 2025.
Introduced an AI-assisted development workflow for a six-person engineering team, combining reusable Claude Code conventions with automated review in GitHub Actions. The new workflow shortened the Dev-to-Test cycle by 30%.
ABOUT
I work across the full product path: shaping the problem with product teams, designing the system, shipping the interface, and making production behavior measurable.
I try most new AI tooling outside of work first, as prototypes I expect to throw away. Very little of it earns a place in the product — the value is knowing which part would, before the team commits to it.
CONTACT