Computer Science student at USC. I build full-stack tools that turn raw, messy data into ranked, actionable leads.

A situational awareness dashboard that visualizes live air traffic, geopolitical incidents, and world news on an interactive 3D globe. Aggregates data from OpenSky Network and GDELT, filters thousands of raw events into meaningful incidents using a four-stage pipeline (severity scoring, CAMEO classification, geographic deduplication via PostGIS, and recency filtering), and renders everything on a Globe.gl visualization.

Tech: Java, Spring Boot, React, PostgreSQL + PostGIS, Globe.gl, Docker, REST API, JUnit/Mockito, Jest

A real-time pipeline that turns a noisy stream of public-record signals (pre-foreclosure notices, tax delinquencies, probate filings) into ranked, explainable leads, built on synthetic data. Fuzzy entity resolution (Jaro-Winkler) merges messy records into homeowner identities at precision/recall/F1 = 1.0 on a fixed-seed benchmark, a rules-based scorer assigns hot/warm/cold tiers with plain-language reasons, and the moment a lead goes hot it streams live to a React dashboard over Server-Sent Events. Benchmarked at ~2,700 signals/sec with p95 signal-to-lead latency under 6ms, backed by 177 JUnit + 19 Jest tests at a 90% coverage gate. Optional Claude-generated "why this lead" explanations via the Anthropic API.

Tech: Java, Spring Boot, React, Server-Sent Events, Apache Commons Text (Jaro-Winkler), Maven, JUnit/Mockito/AssertJ, JaCoCo, Jest

Demo: 10-min walkthrough

A team-built web application where users create meals from recipe searches, host events with guest invitations, and generate deduplicated shopping lists. Built with strict TDD (100% test coverage), Cucumber/Selenium E2E testing, full GitFlow, and bi-weekly stakeholder reviews following SCRUM.

Tech: Java, Spring Boot, React, Tailwind CSS, JUnit/Mockito, Cucumber, Selenium, Jest

A full-stack rideshare app for verified USC students to post rides, discover matches, and coordinate travel. Features a multi-factor match scoring algorithm (time proximity, cost-split alignment, passenger capacity, origin similarity) that ranks compatible rides on a 0–1 scale, real-time WebSocket chat (STOMP + SockJS) between matched riders, Google OAuth 2.0 with server-side token verification and @usc.edu enforcement, and a scheduled background service for automatic ride expiration.

Tech: Java, Spring Boot, React, Tailwind CSS, PostgreSQL, Supabase, WebSocket/STOMP, JWT, Google OAuth 2.0

A Unity game prototype built around polarity interactions and momentum-based mechanics. Implemented a van der Waals-inspired attraction/repulsion model with distance-dependent force curves and minimum-distance clamping for "slingshot" projectile redirection. Added curved pursuit steering and telegraphed timed detonation with AoE destructible damage.

Tech: Unity, C#, Physics2D, gameplay systems design, rapid prototyping

Languages: Java, C++, JavaScript, TypeScript, Python, C#
Frameworks & Tools: Spring Boot, React, Tailwind CSS, PostgreSQL, PostGIS, Supabase, Unity, Docker, Maven, Git, REST APIs, Globe.gl
Auth & Security: Google OAuth 2.0, JWT, HTTPS/SSL
Testing & Engineering: JUnit 5, Mockito, Jest, Cucumber, Selenium, TDD, 100% coverage enforcement, GitFlow, SCRUM

• LinkedIn: linkedin.com/in/anthonyoliv
• Email: tolivare@usc.edu