Advisor: Prof. Yue Zhao - NLP Anomaly Detection Toolkit and Benchmarking Development | LLM, AutoML, NLP, Anomaly Detection, PyTorch

• Created a robust benchmarking system for evaluating algorithmic efficacy with over 40 PyOD library O.D. algorithms and 3 end2end algorithms across 20 datasets to facilitate comprehensive performance analysis and comparison.
• Redesigned NLP datasets into a unified format for anomaly detection, addressing a field gap, and developed a system to automatically recommend optimal AD algorithms based on benchmarks.
• Co-first authored a systematic review on AI-aided digital twin design, analyzing how machine learning enhances digital twins and their applications across multiple domains.
• Constructed causal reasoning datasets, fine-tuned LLMs, and evaluated using hallucination benchmarks.
• Investigated generative AI techniques to tackle Anomaly and Out-of-Distribution Detection problems and explored the capabilities of LLMs in politics.

Advisor: Dr. Yue Cheng - Effective Machine Learning Systems Development | LLM Inference, LLM compression

• Researched an automated method to generate the optimal LLM inference compression algorithm for user-specific tasks.

Advisor: Dr. Jiechao Gao - Federated Learning System Development | Federated Learning, PyTorch, Python, Spatio-temporal data

• Developed the FedMetaMed, integrating federated learning and meta-learning to enhance personalized medication strategies across distributed healthcare systems, improving model adaptability and privacy preservation.
• Single-handedly developed FedLDR, a brand new federated learning algorithm that employs GCN to enhance spatio-temporal data analysis through local data integration and node-centric optimization. = FedMetaMed and FedLDR[were accepted by BIBM and ICDM respectively, and I gave talks on these research.

Advisor: Dr. Jiechao Gao, Prof. Brad Campbell - Hierarchical Federated Learning System Development | Hierarchical Federated Learning, PyTorch, Python, Sparse Network

• Independently developed H-FedSN pushes the boundaries of IoT with a unique approach that uses masking techniques to train a sparse network, enhancing personalization through client-based transfer learning. Applied to non-IID IoT datasets, it achieves high accuracy and boosts communication efficiency by at least 58x.
• Solely developed and integrated innovative federated learning algorithms—FedAvg, FedCAMS, FedPer, PerFedAvg, and FedRS—into a hierarchical framework to optimally benchmark against H-FedSN.

Advisor: Dr. Iordanis Fostiropoulos - Distributed ML Execution Framework Development | MLSys, AutoML, Ray, Docker, Github Action, PyTorch, Pytest

• Contributed to “Ablator”, an open-source Deep Learning framework used by 40+ USC researchers for horizontal scaling of ablation experiments and hyperparameter tuning, encompassing 70 pull requests.
• Implemented distributed experiment execution with Ray, managed open-source projects, set up CI pipelines via GitHub Actions, oversaw release management and version control, and authored pytest unit tests with 97% coverage.
• Solely launched ‘python-rclone’ on PyPI, a Python API for RClone that streamlines cloud data synchronization for ‘Ablator’, removing pre-installation requirements and enabling automatic binary selection (python-rclone).

Advisor: Prof. Guoquan Wu - Automated Testing Platform Development | Docker, Node.js, JSON, Vue.js, RobotFramework

• Contributed to the R&D of a web-based automated testing tool using Record and Playback technology, significantly enhancing test case management by enabling streamlined recording, editing, execution, analysis, and result generation. This implementation boosted end-to-end testing efficiency by 300% and saved over 15 hours per week.
• Independently developed a script parser using Node.js that converts user actions recorded in JSON format into executable Robot Framework and Selenium scripts, enabling the replay and repeated execution of these user actions.
• Single-handedly created innovative UI components using Vue.js and AceEditor, orchestrated the optimal containerization of the program with Docker, and automated the DevOps pipeline to maximize development efficiency.

SaaS Platform Development {Demo} | Kubernetes, Docker, Go, CRD, Operator-SDK, Helm3, Prometheus, Grafana

• Developed “RocketMQ as a Service”, akin to “RabbitMQ as a Service” in AWS Marketplace, offering fully managed SaaS-based RocketMQ clusters, increasing 100% creation speed and saving 10+ hours/week in manual operations.
• Utilized Operator SDK to build a Kubernetes-based RocketMQ Operator and CRD automating lifecycle management.
• Employed Helm3 to package RocketMQ’s components into Helm charts, simplifying Kubernetes deployment.
• Implemented Prometheus and Grafana for real-time monitoring of critical service metrics and node health.
• Automated workflows, including unit tests, image builds, and Helm3 Chart updates, via GitLab CI/CD.
• Researched and evaluated container runtimes (sysbox, crun, youki) for suitability as replacements for Docker in the SaaS platform, ensuring CRI-O compliance and robust community support.
• Optimized a machine learning training pipeline using GPUs on Kubernetes for enhanced computational efficiency.

Text Translation Platform Development | Java, SpringBoot, Spring, Java Persistence API, Maven, Nginx, MySQL, Git

• Developed an AI document translation system with Java/Spring/Maven, independently created a PDF/XML parsing module attracting 30,000 MAUs, used Nginx for reverse proxy, and managed version control with Git.