Software Engineer | PostgreSQL Internals, AI and Backends

Worked on AgensGraph core and AI systems remotely, improving PostgreSQL-based graph query execution, indexing, data loading, security, and backup/upgrade compatibility. Replaced sequential scans with B-tree index probes in path expansion, reducing a two-hop traversal on the 30.6M-edge Pokec graph from a timeout exceeding 20 seconds to approximately 6 ms warm.

Improved DETACH DELETE performance by approximately 30x in local tests, rebuilt CSV loading with PostgreSQL COPY and heap_multi_insert for 15–20% better performance, and implemented GQL query chaining, variable bindings, correlated Cypher subqueries, list comprehensions, native graph types, and default graph indexes.

Built AgensGraph AI integrations and MCP servers for Cypher, memory, and data modeling. Developed LangChain, LangGraph, LlamaIndex, Cognee, and LightRAG adapters; implemented secure read-only LLM query execution; and created execution-graded text-to-Cypher evaluations that improved accuracy from 60% to 73%.

Served as an Apache AGE committer since 2023 and PMC member since 2025, contributing to the PostgreSQL graph database project. Reviewed more than 290 pull requests, merged more than 180 personal contributions, helped ship releases, and maintained CI across PostgreSQL versions.

Implemented and maintained functionality involving graph labels, pg_dump, pg_restore, pg_upgrade, constraints, row-level security, PostgreSQL extension functions, operators, planner and executor behavior, and graph query performance.

Co-founded a software agency and led delivery for courier-aggregation and finance/tax applications.

Contributed to product development and backend implementation, supporting client application delivery from technical planning through implementation.

Completed a Bachelor of Software Engineering degree.