Business Analytics & AI student at UT Dallas.

I work at the application layer — not training models, but making them dependable inside systems people actually use.

Instead of a vague match percentage, it maps every job requirement to the specific evidence that proves it — and quotes the source. pgvector semantic search narrows the candidates, then an LLM validates each one with a confidence band and a full audit trail.

Next.js · PostgreSQL + pgvector · embeddings · OpenAI — live demo

Three stages, built during my internship at Global Experience Specialists. Each one feeds the next. Presented to the CIO.

1. TradeShow-Calendar-Cleaner — turns messy trade-show HTML into clean structured data, cutting data prep time ~90%
2. Market intelligence dashboard — ranks 4,000+ global trade shows by ROI potential (Power BI, internal)
3. AI outreach agent — finds the event organizer's primary contact and drafts a tailored email, then stops for human approval before anything sends. Orchestrated with LangGraph. (private)

Working with: RAG · vector search · embeddings · LLM APIs · LangGraph orchestration · human-in-the-loop systems · Python · TypeScript · SQL · Power BI · ETL pipelines

📍 Dallas, TX · LinkedIn · mekylesiddiqi@gmail.com