At TresVista Analytics, I build production-ready LLM applications and multi-agent systems for venture capital investment intelligence using AWS Bedrock, AgentCore, Claude, and LangGraph. I developed a five-phase pipeline that extracts people, therapy, and finance information from healthcare pitch decks.
I also built an investment-memo workflow and event-driven automations for deal classification, document extraction, and company matching. LLM-based entity resolution improved company-name reconciliation accuracy by more than 30% and eliminated over 20% of duplicate records, while AI-assisted automation reduced manual deal-triage time by approximately 40%.
