Solo build for NNE, in collaboration with domain engineers across 28 disciplines
Fullstack Developer — graph modeling, AI assistant, and visualization layer
Neo4j, Next.js, D3.js, LLM-to-Cypher translation, Prompt Engineering
NNE's Engineering Model — a List of Deliverables spanning 1,800+ entries across 28 disciplines, a 6-level hierarchy, and 5 project phases — lived in a flat Excel spreadsheet. Navigating cross-discipline dependencies or answering questions like “what feeds into this deliverable” meant manual lookups that flat spreadsheets, and even traditional RAG, couldn't support.
I built a proof-of-concept that models disciplines, phases, activities, and deliverables as typed nodes in a Neo4j knowledge graph, connected by relationships like CONTAINS, FEEDS INTO, EVOLVES TO, and BELONGS TO — enabling multi-hop dependency traversal across the entire model.
On top of the graph, I built an AI conversational assistant that translates natural-language questions directly into graph queries, returning multi-hop answers (cross-discipline impact chains, phase-transition dependencies) and structured aggregations (counts, groupings) straight from the graph.
I designed and built the graph schema, the natural-language-to-graph-query assistant, and the full-stack visual layer end-to-end using Next.js and D3.js — delivering three integrated, graph-backed views: an interactive Hierarchy Navigator, a Cross-Discipline Dependency Graph, and a Phase Timeline, each driven by live graph queries.