Three roles. Production systems. Real users.

Member management platform. in progress Centralizing operations across the organization — application review, project assignment, alumni records — into one system. Replacing six disconnected spreadsheets and three Notion bases.

Finance automation. in progressEliminating recurring manual overhead — reimbursements, sponsor invoicing, budget rollups — by codifying the board's monthly workflow into a small, auditable pipeline.

Agentic AI chatbot on AWS. Built and deployed to production with a multi-agent architecture leveraging a RAG pipeline with source citations. Full-stack: frontend, backend, APIs. Presented by senior management at an international Deloitte AI conference in Japan to acquire new enterprise clients.

Lead prioritization model. Trained an ML model for customer lead ranking in automotive aftersales — measurably improving efficiency in lead management for a major OEM client.

Selected for Deloitte's Top 10% Talent Pool — a closed program for high-potential interns based on outstanding internship performance and partner feedback.

MCP server for LLM context. Designed and deployed an MCP server for structured LLM context delivery — caching at the boundary cut API latency by half, improved context reliability, and lowered model-call cost across internal AI features.

Natural-language filtering.Designed, implemented, and shipped to production an AI-powered free-text search and filter system that maps natural language into structured filter state — live on Check24's main consumer surface.