CS & Engineering @ UCLA · AI / SWE · AI Solutions Intern @ Sagitec

I build agentic AI: the kind where an LLM does reliable work inside a real engineering pipeline, not a demo that falls apart on the second try. Most of what I care about is the unglamorous part, grounding, validation, and keeping the model from making things up.

• Sagitec (AI Solutions Intern) — Building the coding and system-test agents in a gated, human-in-the-loop multi-agent SDLC pipeline that generates against a proprietary low-code language no base model has ever seen. Tool-calling harness, Lucene-backed RAG for grounding, MCP integration, and a ReAct loop I cut per-task token usage on by 90%+.
• OpenSrcer — An open-source agentic framework on the Model Context Protocol (MCP) that standardizes tool-calling and system-level actions for autonomous agents. Won Best Use of AI/ML at CitrusHack.
• Agentic +EV analytics platform — A multi-agent LangGraph pipeline over PostgreSQL that models player-prop outcomes and flags +EV signals, built like a trading desk. The LLM orchestrates queries; it never invents a number.
• Systematic backtesting framework — A reproducible research engine for multi-asset strategies with real transaction-cost modeling and rigorous statistical validation (walk-forward, out-of-sample, permutation tests).
• Daily Bruin — Performance work and features across the web stack for one of the nation's largest student-run papers (176k+ monthly readers).

• OpenSrcer — MCP agent framework; Best Use of AI/ML at CitrusHack (solo, 300+ participants).
• WealthWise — Congressional App Challenge winner, presented at the U.S. Capitol.
• Open source — recently merged a fix into worldmonitor (82k+ stars).

• High School Valedictorian, now doing CS & Engineering at UCLA.
• Lift on an upper/lower split; UCLA's hills double as extra leg day.
• Watch Nasdaq order flow more than is probably healthy.
• Still fold origami. Sometimes the best systems don't need electricity.
• SF 49ers and Sacramento Kings.

LinkedIn · Email · Los Angeles, CA