Rick L.

I am a machine learning engineer specializing in LLMs, AI agents, and data pipelines. I build working prototypes, evaluate models, and deliver documented code.

Research: First author of HypTip, “Hyperbolic Language Models: From Zipf to Compute-Optimal Scaling,” accepted to NeurIPS 2026. The work studies hyperbolic LM output layers through pretraining experiments, scaling-law analysis, and benchmark evaluation.

Industry experience:

• Tencent CDG Advertising, full-time Machine Learning Engineer (2026-present): Reworked a RAG knowledge base into an LLM Wiki using GLM-5.2; optimized database-table retrieval; built a self-improving agent with SkillOpt and GEPA; coordinated cross-business data integration and governance.

• Didi, Technical Algorithm Intern: Adapted and integrated the RSS driving-safety framework; validated driving scenarios in Carla; worked with C++, ROS, Docker, and Protobuf.

• Momenta, Machine Learning Intern: Developed PyTorch autoencoder-based deep clustering for hazardous driving scenarios and analyzed autonomous-driving data with Python/SQL.

AIJobTech cofounder: Developed a LangChain agent for job-application workflows, a Python/MySQL backend, and AWS EC2/RDS deployment.

Services: LLM integration and evaluation, AI agent workflows, model training and fine-tuning, data analysis, and ETL pipelines.

Tools: Python, PyTorch, Transformers, LangGraph, LangChain, Pandas, SQL, Spark, AWS.