I work across conventional software engineering and AI-agent application development—building APIs, backend services, data workflows, tool-using agents, evaluation pipelines, and production infrastructure.

• Kaggle Competition Silver Medal — placed 184th of 4,186 teams in AI Agent Security — Multi-Step Tool Attacks, sponsored by OpenAI, Google, and IEEE.
• Tencent Rhino-Bird Open Source Program — Issue Top 3 — completed four issue-practice tasks across OpenTenBase and OpenCloudOS, earned open-source practice certificates, and was named to the official honor roll.
• Academic standing: Top 20% in the Computer Science cohort at Southwest University.
• TOEFL iBT 5.0/6.0 on the 2026 scale — CEFR C1, comparable to 95–106 on the previous scale (ETS score guidance).
• Built and shipped AgentConfigScore — an open-source regression gate for AI coding-agent instructions, published on PyPI and usable in GitHub Actions.
• Contributed merged patches to large open-source systems including .NET Runtime, Apache Airflow, Microsoft PowerToys, Apache Fineract, and HiveMind.

A deterministic tool for detecting regressions in AGENTS.md, CLAUDE.md, Cursor, Copilot, and Gemini instruction files. It provides baseline-aware scoring, line-addressable findings, GitHub code-scanning output, and CI enforcement without runtime dependencies.

• Tencent Rhino-Bird 2026 — completed four open-source tasks across OpenTenBase and OpenCloudOS: Issue #199 through PR #253, Issue #200 through PR #254, an OpenCloudOS Security Skill — PR !86 (completion record), and parallel chroot package testing — PR !10 (completion record); both OpenCloudOS tasks were marked “completed and awarded,” with Issue Top 3 recognition.
• .NET Runtime — runtime metadata and a performance-sensitive API implementation.
• Apache Airflow — safer Kubernetes git-sync configuration.
• Microsoft PowerToys — stable command identity in Command Palette.
• Apache Fineract — frontend test modernization and maker-checker workflow handling.
• HiveMind — message handling and checkpoint management for a multi-agent system.

I am a Computer Science undergraduate at Southwest University, preparing for graduate study in the United States. My focus is the software engineering around AI rather than model training: agent orchestration, tool use, context handling, model and API integration, evaluation, backend systems, data workflows, and production infrastructure.

For conventional SDE work, I build backend and full-stack systems with Java/Spring, Python, TypeScript/React, SQL databases, Docker, automated testing, and CI/CD. Beyond the work shown publicly here, I have built Java backend, AI-agent, and database-backed full-stack systems in private and academic settings. I am comfortable moving between the emerging agent stack and established software stacks—turning ambiguous product requirements into software that can be tested, maintained, and shipped.

_{Build the system. Test the claim.}