Active Software Systems

OBJECTIVE

Automate end-to-end literature review workflows by orchestrating LLM-driven agents, multi-source fetchers, and RAG-based retrieval to enable scalable, reproducible screening, extraction, synthesis, and reporting across research teams.

ENGINEERING HIGHLIGHTS

• Agent-driven pipeline integrating fetchers for PubMed, IEEE, Scopus, and WOS with normalized responses and configurable node-based prompts.
• RAG + vector-store pipeline (chunker, embedder, engine) enabling semantic full-text retrieval and near-real-time document synthesis.
• Modular LLM provider and streaming chat support for interactive review sessions and incremental extraction across large corpora.
• Next.js frontend with workspace management, PDF upload, and React hooks paired with backend state machine and Prisma models for reproducible workflows.