🛠️ Projects

Enterprise AI Integrated System (企业级AI一体机)

• [Role] Product Lead & Founder
• [Highlight] Next-generation AI infrastructure for enterprise digital transformation, innovatively integrating high-performance computing with AI agent technology. Supports private local deployment and low-code workflow agent development. Core philosophy: “Computing as a Service, Intelligence as Productivity.”
• [Architecture] Three-tier decoupled architecture: Computing Layer + Model Layer + Application Layer, enabling rapid deployment and flexible scaling of enterprise-grade AI capabilities.
• [Achievement] Successfully deployed to dozens of enterprises across manufacturing, new energy, finance, and legal sectors. Incubated multiple benchmark applications:
  • AI-assisted CNC Programming System (60% efficiency improvement)
  • Intelligent Power Trading Assistant (1000+ daily trading decisions)
  • Legal Due Diligence Report Assistant (80% time reduction)
  • Intelligent Patent Search Engine (95%+ recall rate)
• [Impact] Headquartered in Shenzhen (R&D) and Wenzhou (National Marketing Center), driving large-scale AI adoption in the real economy.

Alkaid - Huawei Cloud Intelligent Resource Scheduling Platform (华为云瑶光资源调度项目)

• [Role] Team Leader (15-person algorithm team)
• [Scale] Core infrastructure project handling 10+ million daily scheduling requests across 30+ global data centers and 2+ million servers.
• [Highlight] Built an end-to-end intelligent scheduling system with offline training, online inference, and real-time feedback. Developed elastic resource allocation model under multi-objective constraints, integrating LSTM time-series prediction with PPO reinforcement learning, achieving 98.5% scheduling accuracy with second-level decision-making for 100K+ node clusters.
• [Achievement] Improved average fragmentation rate by 30%, saved 6% of servers, extended defragmentation cycle by 50%.
• [Profit] Generated over 200 million RMB in revenue growth for Huawei Cloud IaaS.

AIOps - Huawei Cloud Intelligent Operations Platform (华为云智能运维项目)

• [Role] Project Manager
• [Highlight] Established a 10 million RMB joint laboratory with The Chinese University of Hong Kong (CUHK), with 10+ PhD researchers over 3 years. Built a millisecond-level closed-loop system for anomaly detection, root cause localization, and self-healing decisions.
• [Innovation] Developed “Spatio-temporal Feature Perception + Causal Reasoning Engine” dual-driven framework. Designed GNN-based service topology anomaly propagation model, solving cross-layer (IaaS/PaaS/SaaS) fault correlation analysis challenges.
• [Achievement] Achieved 98.7% multi-dimensional anomaly detection accuracy, reduced root cause localization time to seconds, supporting Huawei Cloud’s 99.99% SLA commitment.
• [Profit] Reduced annual O&M costs by 230 million RMB.

AILeader - One-Stop AI Education Platform (智核苑一站式AI教学平台)

• [Role] Founder
• [Highlight] China’s first one-stop AI education platform for K-12 students (ages 5-18). Integrates three core modules: AI Programming, AI Tutoring, and AI Science & Innovation. Features block-based programming, instruction-based programming, learning assistants, mock exams, AI art creation, and digital life modules.
• [Achievement] Pioneered AI enlightenment curriculum for Chinese youth. The innovative AI course system has been adopted by primary and secondary schools in Guangzhou, becoming a national “AI Science & Innovation” demonstration program.
• [Profit] AI course packages and platform subscriptions have generated over 2 million RMB in revenue. Enrolled 100+ students at the flagship Cuihu Branch in Tianhe District, Guangzhou.

HeTao AI Mini Program (核韬AI小程序)

• [Role] CTO
• [Highlight] Consumer-facing AI generation tool covering AI Q&A, dialogue, painting, portrait generation, and AI picture books.
• [Achievement] Accumulated 500,000 users with 3,000 daily active users. Obtained software copyright. Passed national internet information service algorithm filing review.
• [Profit] Advertising and premium subscriptions exceeded 1 million RMB in revenue.

CoreMI - AI-Assisted Medical Imaging Diagnosis System (AI辅助医疗影像诊断系统)

• [Role] Project Lead
• [Highlight] Developed a multimodal deep learning system for intelligent pulmonary nodule diagnosis at tertiary hospital standards. Covers the full pipeline: CT image preprocessing → lesion detection → benign/malignant analysis → structured report generation.
• [Innovation] Pioneered 3D Multi-scale U-Net++ architecture, integrating DenseNet feature reuse and attention gating mechanisms. Built spatial-density-morphology multi-dimensional feature extraction network, achieving 94.3% F1-score on LIDC-IDRI dataset (surpassing industry benchmark by 7.6 percentage points).
• [Achievement] Achieved 98.2% sensitivity and 96.5% specificity for pulmonary nodule detection. Improved diagnostic efficiency by 5x and reduced misdiagnosis rate by 20%. Passed CFDA Class II medical device certification. Selected as Tsinghua X-lab Outstanding Demonstration Project.
• [Profit] Piloted at a top-tier hospital. Granted one algorithm invention patent and one software copyright.