About
Hi, I'm zhihanz.
I'm a Solution Architect at a stealth company working at the intersection of GenAI, agentic workflows, and data infrastructure. My focus: making AI systems that actually work in production—grounded in real data, reliably orchestrated, and built to scale.
What I'm Building Now
The GenAI wave has created an urgent need to connect LLMs to structured data, orchestrate complex multi-step workflows, and build systems that are observable, cost-efficient, and don't hallucinate your business into the ground. That's where I live.
I spend my days thinking about agentic architectures, RAG and data pipelines, and workflow orchestration—turning brittle prompt chains into resilient, production-grade systems.
The Foundation
This work is built on a decade of database and distributed systems experience.
I co-founded Databend and helped build Databend—a cloud-native, Rust-based query engine. I took it from first commit to Fortune 500 customers, learning that building a database is hard, but building a business around one is harder.
Before that, I worked on infrastructure at Google (Istio/Service Mesh) and Nebula Graph (distributed graph DB). I've debugged Kubernetes networking race conditions, implemented LLVM-JIT query engines, and served as a community release manager for Istio.
I also published research at ECCV 2020 on neural networks for image manipulation detection—an early adopter of attention mechanisms before ChatGPT made them mainstream.
What I Write About
- GenAI & Agents – Patterns for building reliable agentic systems, RAG architectures, and LLM orchestration.
- Data Infrastructure – Query engines, storage layers, and the plumbing that makes AI possible.
- Rust & Systems Programming – Including how we made our SQL parser 33x faster.
Offline
I live in the San Francisco Bay Area. When I'm not architecting workflows or profiling latency, I'm vibe coding—letting AI write the first draft while I steer.
Find me
Feel free to reach out!