.png){kind=logo}
Huihai Wang
Mid-level Applied AI Engineer specializing in knowledge graphs, GraphRAG, and urban mobility
Connect with Huihai
Huihai already has a relationship with Reval, so a warm intro from us gets a much better response than cold outreach.
Recommended
Already have an account?
About
ML/NLP practitioner focused on knowledge-graph-based retrieval for LLM question answering, including an urban/autonomous-vehicle decision-making use case. Built a hierarchical GraphRAG + vector database system and an entity-resolution pipeline that blends spatial and semantic similarity, validated using LLM-generated synthetic datasets; uses Python tooling like RDFLib, GraphDB, OpenAI APIs, and LangChain.
Find your next great hire
Our AI agents source, screen, and vet candidates for your open roles. Get qualified candidates within 48 hours.
Key Strengths
- Built an urban-domain knowledge graph to improve LLM reasoning for urban/autonomous-vehicle decision-making QA
- Designed hierarchical GraphRAG + vector-store retrieval over separated KG subgraphs (roads, accidents, POIs)
- Entity resolution approach combining spatial filtering with semantic vector similarity on names/descriptions
- Created synthetic validation datasets by rewriting entities/descriptions with LLMs and adding hard negatives
- Pragmatic optimization of matching thresholds and cost/time via staged filtering (spatial + traditional matching first)
- Modular, extensible Python workflow design with ontology interfaces for future dataset expansion
Like what you see? We'll introduce you to Huihai directly.
Experience
Education
Skills
Languages
Publications
7 publications
Memberships
Interested in Huihai?
We'll personally introduce you - no strings attached.
Build your dream team with Reval
Our AI agents source, screen, and vet candidates for your open roles. Get qualified, high-intent candidates on your desk within 48 hours.
Discover more candidates like Huihai
Search across thousands of pre-screened, high-quality, high-intent candidates on Reval.
Search Talent