may 2025 — present
Software Engineering Intern, AI & Full Stack @ BiblioCommons
Building AI tooling and full-stack features for library software used by 20M+ patrons, from a company-wide RAG platform to production bug fixes.
- Architected a RAG platform (Spring AI, pgvector) over 50,000+ documents, reaching 70% company-wide adoption and cutting retrieval time 80%.
- Built an agentic AI SRE assistant over Sumo Logic, Prometheus, Checkmk and Confluence/Jira; cut incident resolution time 50%+ and doubled the on-call pool.
- Cut p95 latency up to 50% on a REST API serving 20M+ users by optimizing database queries and business logic.
- Raised RAG retrieval accuracy to 95% on a labelled eval set with retrieval agents that re-query when context is missing.
- Closed 100+ production bugs for 250+ library customers and shipped WCAG 2.2-compliant React features.