Ruizhi Chen

Achievements: ICPC Team

Coursework: Data Structure and Algorithms, Discrete Math, Linear Algebra, Calculus I, II, Computer Organization

• Engineered a state-of-the-art browser agent that achieved a 72% success rate on the OSWorld benchmark, utilizing Chrome DevTools Protocol (CDP) for deep-system injection and real-time DOM manipulation.
• Developed a proprietary domain-specific language (Simulung) to streamline agentic workflows, reducing instruction latency and improving task execution accuracy across complex web environments.

• Researched statistical arbitrage strategies with feature pipelines and walk-forward validation.
• Built machine learning models for market prediction; explored microstructure features for low-latency trading signals.
• Prototyped high-frequency trading algorithms on historical order book data using AWS, improving algorithm efficiency and readiness for implementation.

• Built and deployed production game systems using FastAPI (Python) and Next.js as part of Ambitus AI's client delivery contract, creating scalable training environments for AI model evaluation.
• Implemented hash-based deterministic randomization with configurable seeds to generate reproducible yet diverse training datasets, ensuring consistent CNN model evaluation across experiments.
• Delivered production-ready code with comprehensive game state management, API endpoints, and validation logic for systematic AI behavioral assessment.

• Migrated a large-scale React.js company site, pre-rendered via a Static Site Generator (SSG), to Cloudflare Pages, cutting annual hosting costs by 97% (from $884 to $20).
• Monitored high traffic volume (~5.6M requests/mo) using Cloudflare Analytics, focusing on the cache-hit ratio for the site's core JavaScript bundle and static assets to identify critical performance bottlenecks.
• Improved median site load time by 87% (4.1s to 0.53s) by implementing CDN caching, configuring optimized cache-control headers, and leveraging JavaScript performance techniques like code splitting and lazy loading of React components.

• Taught 100+ students in AP CS A (Java) and USACO Bronze–Silver competitive programming.
• Hosted office hours for 100+ students, mentored and evaluated other TAs.

• Engineered a massively parallel trading simulation platform leveraging Python multithreading to support 200+ concurrent traders executing 50,000+ orders daily with <100ms latency.
• Architected a multi-threaded event-driven system with thread-pool executors processing 2.5M+ data points per hour across independent market feed handlers.
• Built a real-time dashboard with dedicated worker threads for each user session, enabling sub-50ms data synchronization across 200+ concurrent WebSocket connections.

• Enabled elderly and casual players to digitize physical games, lowering barriers to analysis and making advanced chess tools accessible to broader audiences.
• Built real-time chessboard detection (CNN, YOLOv8, Lucas-Kanade optical flow, KD-Tree filtering). Generated Forsyth-Edwards Notation (FEN) and integrated with Stockfish for automated move analysis.

• Co-founded an international competition reaching 350+ participants from 40+ countries, lowering entry barriers to advanced CS education.
• Managed $50K prize pool; secured sponsors (JetBrains, Wolfram|Alpha).