Yanan Gong

Phone: 434-466-5462
Email: gongyn3@gmail.com
LinkedIn: linkedin.com/in/yanan-gong-yg
GitHub: github.com/Yanan-Gong

Technical Skills

Programming & Databases: Python, SQL, Neo4j
Frameworks & Tools: PyTorch, Hugging Face, Scikit-learn, PySpark, Hadoop, Git, ETL
Cloud & Deployment: Azure, AWS, MLflow, Kubernetes, Airflow, Docker, Databricks, MLOps, CI/CD
Highlights: LLM, NLP, GenAI, RLHF, Foundation Models, A/B Testing, Machine Learning, Deep Learning

Education

M.S. in Statistics | University of Virginia, Charlottesville, VA | GPA-3.9/4.0 | Aug 2019 - May 2021

B.S. in Applied Statistics | Sun Yat-sen University, Guangdong, China | GPA-3.5/4.0 | Sep 2014 - Aug 2018

Work Experience

Senior DS/MLE | Air-Conditioning, Heating, and Refrigeration Institute, Arlington, VA | Aug 2021 - Present

AI Selection Copilot & MLOps (BI Tool):

  • Initiated and led development of a GPT-4-based multi-agent AI Copilot using Retrieval-Augmented Generation (RAG), automating mechanical audits and reducing engineering workload by 80%
  • Wrote production-grade ML pipelines integrated with CI/CD workflows, deployed via Docker and Kubernetes for scalable and reproducible model deployment
  • Developed RESTful APIs for low-latency ML inference services, enabling real-time engineering decisions with sub-second response time
  • Implemented model versioning and monitoring using MLflow and Airflow, ensuring continuous delivery and rollback support in production
  • Leveraged LLM as an evaluator to assess output accuracy and relevance, applying conceptual fine-tuning to improve robustness

Anomaly/Fraud Detection: Developed and deployed an end-to-end XGBoost to identify anomalies across 10 million HVAC ratings in real-time, reducing operating costs by $500,000 annually. Achieved a recall of 97% with precision of 85% and an F1-score of 90%, ensuring critical anomalies are captured.

Cross-functional Collaboration: Led cross-functional collaboration with engineers, legal, and policy teams, engaging with multiple stakeholders to ensure cohesive user experiences and industry compliance.

Leadership: Mentored multiple interns on project completion; Connected new hires with company culture.

Research Assistant | Right Legal & University of Alberta, Edmonton, Canada | Sep 2017 - May 2018

  • Implemented a multi-label text classification system for historical legal documents, preprocessed featured words by Word2Vec to convert professional legal terms into plain language
  • Built and deployed a Transformer based model achieving 94% accuracy and conducted sentiment analysis to enhance the website’s matching algorithms, improving client-lawyer pairing efficiency

Projects

Mock Interview AI Agent - LLM Agents Hackathon | Python, Flask, AWS | Dec 2024

  • Led, designed, and deployed a scalable AI-powered mock interview platform leveraging GPT-4, RAG, and multi-agent architectures. Delivered role-specific questions, industry-tailored scenarios, and real-time AI feedback, reducing user costs by 80%
  • Implemented Reasoning and Action prompt engineering and LLM-based evaluations, deploying on Kubernetes with auto-scaling, supporting 1000 QPS and future scalability