Built and evaluated incremental ingestion for semantic search over SysML data.
- Built an ingestion and semantic search pipeline for company-provided SysML data using two embedding models and Qdrant.
- Implemented change-aware updates that detect data changes and re-embed only affected content, avoiding redundant computation.
- Evaluated the system through five evaluations and repeated benchmark runs, comparing update speedups, semantic correctness, and interruption recovery.




