Hi, I'm Zoey Zhang 👋
AI Engineer building production RAG systems, agent architectures, and GenAI applications — with research depth in Knowledge Graphs and NLP.
What I Build
RAG Systems
Production retrieval pipelines with measured quality: p95 228ms retrieve latency, recall@5 0.933, systematic reranking evaluation (+7-11pp improvement).
Agent Architectures
Production agent engines with L0/L1/L2 routing, tool integration, MCP protocols, and cost tracking down to 6.16e-06 USD per query.
Eval-Driven Development
Every feature measured: latency, recall, cost, accuracy. Honest about what works and what doesn't — because production demands truth.
Data & Governance
Structured data pipelines, ontology extraction, knowledge graph construction. AI disclosure, prompt tracking, and transparent cost accounting.
Scalability
Service-oriented architecture with modular, independently deployable components. RAG services consumed by multiple products without tight coupling.
Performance
Optimized retrieval with sub-second latency, intelligent routing to minimize LLM calls, and cost-aware design from architecture to implementation.
Featured Projects
End-to-end AI products — from research prototype to production deployment
GraphRAG Module
Knowledge graph + RAG module for the company's RAG system. Led ontology extraction, systematic reranking evaluation (+7-11pp over baseline, <2s latency), and resolved critical chunk retrieval bug (+20-80% recall).
SceneFlow Video Agent
End-to-end video agent with 7000+ line engine, L0/L1/L2 routing, MCP tool integration, image/video generation (Seedream/Seedance), and cost tracking per query tier.
IntelliScribe
Knowledge-ingest + writing application. Migrated from Chroma to studio HTTP retrieve (p95 228ms, 7.67 rps), SSE token streaming, AI disclosure on health endpoints.
NLQxform (KGQA)
Language model-based question to SPARQL transformer. CEUR 2023 workshop paper + SIGIR 2025 demo. Research in knowledge graph question answering with interactive scholarly QA system.
Skills & Expertise
Languages
Python, TypeScript, Java, SQL
AI / ML
RAG, LLMs, Agents, KG, NLP, Prompt Engineering
Frameworks
FastAPI, Next.js, React, PyTorch, LangGraph
Infrastructure
Docker, PostgreSQL, Neo4j, AWS, HPC