SOFTWARE ENGINEER / UC SAN DIEGO
I’m a computer science student who enjoys building backend systems, machine learning applications, and full-stack products. I like taking ideas from the first sketch to working, tested software.
WHAT I CARE ABOUT
Reliable systems, thoughtful tradeoffs, and products that are useful in practice.
01 / SELECTED WORK
EDGE AI / IOT / FULL STACK
A privacy-first occupancy system that replaces camera monitoring with an 8×8 thermal sensor and on-device inference.
I trained and quantized a TensorFlow Lite Micro model, deployed it to an ESP32, and built the MQTT, FastAPI, MySQL, OAuth, and WebSocket pipeline behind a live monitoring dashboard.
ENVIRONMENTAL ML / DATA VISUALIZATION
A California coastal monitoring system that estimates harmful algal bloom risk up to seven days ahead from buoy and historical bloom data.
The project combines NOAA NDBC sensor data, CalHABMAP observations, feature engineering, Random Forest and XGBoost models, and an interactive station dashboard for weekly risk monitoring.
MULTI-AGENT SYSTEMS / HACKATHON WINNER
A coordinated AI travel planner that turns one request into a ranked, complete itinerary—without the usual tab-sprawl.
A supervisor coordinates specialized flight, hotel, activity, and review agents across a LangGraph workflow, with live search and deterministic price-and-rating filters before final synthesis.
APPLIED ML / DATA PIPELINES
A precision-biased product reconciliation pipeline for finding credible Walmart-to-Wegmans counterparts at catalog scale.
The two-pass matcher combines brand and size normalization, TF-IDF retrieval, deterministic variant gates, and an LLM judge—then documents the precision/recall tradeoffs behind every threshold.
HEALTHCARE AI / FULL STACK
An AI-assisted engine for resolving conflicting medication records and surfacing patient-data quality issues across EHR sources.
The application turns structured clinical records into explainable reconciliation decisions, with schema-constrained model output, source reliability rules, rate-limit handling, and end-to-end MIMIC tests.
02 / EXPERIENCE
Built a production cold-outreach workspace for 21,000 contacts across 8,700 HOAs, increasing outbound calling volume by 17% with server-side filtering, workflow tracking, and an embedded RingCentral dialer. Built an n8n and Gemini vendor-extraction pipeline with citation-backed verification and PII safeguards after auditing 61 client communities.
Built data pipelines for an LLM-based interior-design planner, processing 4,964 design-skill records from 181 videos across 36 room categories. Developed an LLM-assisted Unreal Engine pipeline using a 99%-accurate computer vision model and icon matching to automate virtual film setup.
Shipped Azure Blob Storage as a complete backend provider across upload, download, delete, copy, and connect-test operations—plus validation, React setup, registry, and database migration changes.
Merged PR #328 ↗Selected from 4,000+ applicants for a year-long AI program and built a fairness-aware income classifier on 32,000+ U.S. Census records.
Helped 60+ students strengthen debugging, data structures, calculus, and independent problem-solving through one-to-one instruction.
03 / PROFILE
I care about systems that are technically rigorous, explainable, and useful in the messy conditions outside a notebook.
EDUCATION
B.S. Computer Science · Expected December 2027
GPA 3.98 / 4.0
CORE TOOLKIT
Python, C++, TypeScript, JavaScript, SQL, FastAPI, React, TensorFlow Lite, scikit-learn, Docker, MySQL, MQTT, LangGraph
RECOGNITION
SanD Hacks AGNTCY Track Winner
ECE140B Demo Day · 2nd Place
BASTA Code2Career Fellow